From 1224a93ada1fdccacd186c5369ccca16da10399d Mon Sep 17 00:00:00 2001
From: Arav Agarwal <aravagar@umich.edu>
Date: Thu, 28 May 2026 11:40:21 -0400
Subject: [PATCH] Use claude to update test directories and structure checks

---
 .../submissions/builder.py                    | 119 ++++++----
 .../submissions/github.py                     | 104 +++++----
 src/submission_checker/checker.py             | 162 +++++++------
 src/submission_checker/models/__init__.py     |   6 +-
 .../models/aggregate/context.py               |  40 ++--
 src/submission_checker/models/structure.py    | 168 ++++++--------
 .../llama3-70b/r16}/accuracy/accuracy.txt     |   0
 .../r16}/accuracy/accuracy_result.json        |   0
 .../r16}/mlperf_endpoints_log_detail.json     |   0
 .../r16}/mlperf_endpoints_log_summary.json    |   0
 .../llama3-70b/r16/point.yaml}                |   0
 .../bad_system/llama3-70b/r16/report.txt      |   1 +
 .../llama3-70b/r16/run_metadata.json          |   5 +
 .../llama3-70b/r16/src/vllm}/.gitkeep         |   0
 .../llama3-70b/r38/accuracy/accuracy.txt      |   1 +
 .../r38/accuracy/accuracy_result.json         |   1 +
 .../r38}/mlperf_endpoints_log_detail.json     |   0
 .../r38}/mlperf_endpoints_log_summary.json    |   0
 .../llama3-70b/r38/point.yaml}                |   0
 .../bad_system/llama3-70b/r38/report.txt      |   1 +
 .../llama3-70b/r38/run_metadata.json          |   5 +
 .../llama3-70b/r38/src/vllm/.gitkeep          |   0
 .../llama3-70b/r88/accuracy/accuracy.txt      |   1 +
 .../r88/accuracy/accuracy_result.json         |   1 +
 .../r88}/mlperf_endpoints_log_detail.json     |   0
 .../r88}/mlperf_endpoints_log_summary.json    |   0
 .../llama3-70b/r88/point.yaml}                |   0
 .../bad_system/llama3-70b/r88/report.txt      |   1 +
 .../llama3-70b/r88/run_metadata.json          |   5 +
 .../llama3-70b/r88/src/vllm/.gitkeep          |   0
 .../llama3-70b/sweep_distributions.csv        |   1 +
 .../bad_system/llama3-70b/sweep_summary.csv   |   1 +
 .../system_desc.json}                         |   0
 .../gpt-oss-120b/r1024}/accuracy/accuracy.txt |   0
 .../r1024}/accuracy/accuracy_result.json      |   0
 .../r1024}/mlperf_endpoints_log_detail.json   |   0
 .../r1024}/mlperf_endpoints_log_summary.json  |   0
 .../gpt-oss-120b/r1024/point.yaml}            |   0
 .../gpt-oss-120b/r1024/report.txt             |   1 +
 .../gpt-oss-120b/r1024/run_metadata.json      |   5 +
 .../gpt-oss-120b/r1024/src/vllm/.gitkeep      |   0
 .../gpt-oss-120b/r128}/accuracy/accuracy.txt  |   0
 .../r128}/accuracy/accuracy_result.json       |   0
 .../r128}/mlperf_endpoints_log_detail.json    |   0
 .../r128}/mlperf_endpoints_log_summary.json   |   0
 .../gpt-oss-120b/r128/point.yaml}             |   0
 .../gpt-oss-120b/r128/report.txt              |   1 +
 .../gpt-oss-120b/r128/run_metadata.json       |   5 +
 .../gpt-oss-120b/r128/src/vllm/.gitkeep       |   0
 .../gpt-oss-120b/r16}/accuracy/accuracy.txt   |   0
 .../r16}/accuracy/accuracy_result.json        |   0
 .../r16}/mlperf_endpoints_log_detail.json     |   0
 .../r16}/mlperf_endpoints_log_summary.json    |   0
 .../gpt-oss-120b/r16/point.yaml}              |   0
 .../gpt-oss-120b/r16/report.txt               |   1 +
 .../gpt-oss-120b/r16/run_metadata.json        |   5 +
 .../gpt-oss-120b/r16/src/vllm/.gitkeep        |   0
 .../gpt-oss-120b/r2048}/accuracy/accuracy.txt |   0
 .../r2048}/accuracy/accuracy_result.json      |   0
 .../r2048}/mlperf_endpoints_log_detail.json   |   0
 .../r2048}/mlperf_endpoints_log_summary.json  |   0
 .../gpt-oss-120b/r2048/point.yaml}            |   0
 .../gpt-oss-120b/r2048/report.txt             |   1 +
 .../gpt-oss-120b/r2048/run_metadata.json      |   5 +
 .../gpt-oss-120b/r2048/src/vllm/.gitkeep      |   0
 .../gpt-oss-120b/r4}/accuracy/accuracy.txt    |   0
 .../r4}/accuracy/accuracy_result.json         |   0
 .../r4}/mlperf_endpoints_log_detail.json      |   0
 .../r4}/mlperf_endpoints_log_summary.json     |   0
 .../gpt-oss-120b/r4/point.yaml}               |   0
 .../gpt-oss-120b/r4/report.txt                |   1 +
 .../gpt-oss-120b/r4/run_metadata.json         |   5 +
 .../gpt-oss-120b/r4/src/vllm/.gitkeep         |   0
 .../gpt-oss-120b/r512}/accuracy/accuracy.txt  |   0
 .../r512}/accuracy/accuracy_result.json       |   0
 .../r512}/mlperf_endpoints_log_detail.json    |   0
 .../r512}/mlperf_endpoints_log_summary.json   |   0
 .../gpt-oss-120b/r512/point.yaml}             |   0
 .../gpt-oss-120b/r512/report.txt              |   1 +
 .../gpt-oss-120b/r512/run_metadata.json       |   5 +
 .../gpt-oss-120b/r512/src/vllm/.gitkeep       |   0
 .../gpt-oss-120b/r64}/accuracy/accuracy.txt   |   0
 .../r64}/accuracy/accuracy_result.json        |   0
 .../r64}/mlperf_endpoints_log_detail.json     |   0
 .../r64}/mlperf_endpoints_log_summary.json    |   0
 .../gpt-oss-120b/r64/point.yaml}              |   0
 .../gpt-oss-120b/r64/report.txt               |   1 +
 .../gpt-oss-120b/r64/run_metadata.json        |   5 +
 .../gpt-oss-120b/r64/src/vllm/.gitkeep        |   0
 .../gpt-oss-120b/sweep_distributions.csv      |   1 +
 .../gpt-oss-120b/sweep_summary.csv            |   1 +
 .../system_desc.json}                         |   0
 .../gpt-oss-120b/r1024}/accuracy/accuracy.txt |   0
 .../r1024}/accuracy/accuracy_result.json      |   0
 .../r1024}/mlperf_endpoints_log_detail.json   |   0
 .../r1024}/mlperf_endpoints_log_summary.json  |   0
 .../gpt-oss-120b/r1024/point.yaml}            |   0
 .../gpt-oss-120b/r1024/report.txt             |   1 +
 .../gpt-oss-120b/r1024/run_metadata.json      |   5 +
 .../gpt-oss-120b/r1024/src/vllm/.gitkeep      |   0
 .../gpt-oss-120b/r128}/accuracy/accuracy.txt  |   0
 .../r128}/accuracy/accuracy_result.json       |   0
 .../r128}/mlperf_endpoints_log_detail.json    |   0
 .../r128}/mlperf_endpoints_log_summary.json   |   0
 .../gpt-oss-120b/r128/point.yaml}             |   0
 .../gpt-oss-120b/r128/report.txt              |   1 +
 .../gpt-oss-120b/r128/run_metadata.json       |   5 +
 .../gpt-oss-120b/r128/src/vllm/.gitkeep       |   0
 .../gpt-oss-120b/r16}/accuracy/accuracy.txt   |   0
 .../r16}/accuracy/accuracy_result.json        |   0
 .../r16}/mlperf_endpoints_log_detail.json     |   0
 .../r16}/mlperf_endpoints_log_summary.json    |   0
 .../gpt-oss-120b/r16/point.yaml}              |   0
 .../gpt-oss-120b/r16/report.txt               |   1 +
 .../gpt-oss-120b/r16/run_metadata.json        |   5 +
 .../gpt-oss-120b/r16/src/vllm/.gitkeep        |   0
 .../gpt-oss-120b/r2048/accuracy/accuracy.txt  |   1 +
 .../r2048/accuracy/accuracy_result.json       |   6 +
 .../r2048}/mlperf_endpoints_log_detail.json   |   0
 .../r2048}/mlperf_endpoints_log_summary.json  |   0
 .../gpt-oss-120b/r2048/point.yaml}            |   0
 .../gpt-oss-120b/r2048/report.txt             |   1 +
 .../gpt-oss-120b/r2048/run_metadata.json      |   5 +
 .../gpt-oss-120b/r2048/src/vllm/.gitkeep      |   0
 .../gpt-oss-120b/r4/accuracy/accuracy.txt     |   1 +
 .../r4/accuracy/accuracy_result.json          |   6 +
 .../r4}/mlperf_endpoints_log_detail.json      |   0
 .../r4}/mlperf_endpoints_log_summary.json     |   0
 .../gpt-oss-120b/r4/point.yaml}               |   0
 .../gpt-oss-120b/r4/report.txt                |   1 +
 .../gpt-oss-120b/r4/run_metadata.json         |   5 +
 .../gpt-oss-120b/r4/src/vllm/.gitkeep         |   0
 .../gpt-oss-120b/r512/accuracy/accuracy.txt   |   1 +
 .../r512/accuracy/accuracy_result.json        |   6 +
 .../r512}/mlperf_endpoints_log_detail.json    |   0
 .../r512}/mlperf_endpoints_log_summary.json   |   0
 .../gpt-oss-120b/r512/point.yaml}             |   0
 .../gpt-oss-120b/r512/report.txt              |   1 +
 .../gpt-oss-120b/r512/run_metadata.json       |   5 +
 .../gpt-oss-120b/r512/src/vllm/.gitkeep       |   0
 .../gpt-oss-120b/r64/accuracy/accuracy.txt    |   1 +
 .../r64/accuracy/accuracy_result.json         |   6 +
 .../r64}/mlperf_endpoints_log_detail.json     |   0
 .../r64}/mlperf_endpoints_log_summary.json    |   0
 .../gpt-oss-120b/r64/point.yaml}              |   0
 .../gpt-oss-120b/r64/report.txt               |   1 +
 .../gpt-oss-120b/r64/run_metadata.json        |   5 +
 .../gpt-oss-120b/r64/src/vllm/.gitkeep        |   0
 .../gpt-oss-120b/sweep_distributions.csv      |   1 +
 .../gpt-oss-120b/sweep_summary.csv            |   1 +
 .../system_desc.json}                         |   0
 .../r128/accuracy/accuracy.txt                |   1 +
 .../r128/accuracy/accuracy_result.json        |   6 +
 .../r128}/mlperf_endpoints_log_detail.json    |   0
 .../r128}/mlperf_endpoints_log_summary.json   |   0
 .../qwen3-coder-480b/r128/point.yaml}         |   0
 .../qwen3-coder-480b/r128/report.txt          |   1 +
 .../qwen3-coder-480b/r128/run_metadata.json   |   5 +
 .../qwen3-coder-480b/r128/src/vllm/.gitkeep   |   0
 .../r16/accuracy/accuracy.txt                 |   1 +
 .../r16/accuracy/accuracy_result.json         |   6 +
 .../r16}/mlperf_endpoints_log_detail.json     |   0
 .../r16}/mlperf_endpoints_log_summary.json    |   0
 .../qwen3-coder-480b/r16/point.yaml}          |   0
 .../qwen3-coder-480b/r16/report.txt           |   1 +
 .../qwen3-coder-480b/r16/run_metadata.json    |   5 +
 .../qwen3-coder-480b/r16/src/vllm/.gitkeep    |   0
 .../r256/accuracy/accuracy.txt                |   1 +
 .../r256/accuracy/accuracy_result.json        |   6 +
 .../r256}/mlperf_endpoints_log_detail.json    |   0
 .../r256}/mlperf_endpoints_log_summary.json   |   0
 .../qwen3-coder-480b/r256/point.yaml}         |   0
 .../qwen3-coder-480b/r256/report.txt          |   1 +
 .../qwen3-coder-480b/r256/run_metadata.json   |   5 +
 .../qwen3-coder-480b/r256/src/vllm/.gitkeep   |   0
 .../r32/accuracy/accuracy.txt                 |   1 +
 .../r32/accuracy/accuracy_result.json         |   6 +
 .../r32}/mlperf_endpoints_log_detail.json     |   0
 .../r32}/mlperf_endpoints_log_summary.json    |   0
 .../qwen3-coder-480b/r32/point.yaml}          |   0
 .../qwen3-coder-480b/r32/report.txt           |   1 +
 .../qwen3-coder-480b/r32/run_metadata.json    |   5 +
 .../qwen3-coder-480b/r32/src/vllm/.gitkeep    |   0
 .../r512/accuracy/accuracy.txt                |   1 +
 .../r512/accuracy/accuracy_result.json        |   6 +
 .../r512}/mlperf_endpoints_log_detail.json    |   0
 .../r512}/mlperf_endpoints_log_summary.json   |   0
 .../qwen3-coder-480b/r512/point.yaml}         |   0
 .../qwen3-coder-480b/r512/report.txt          |   1 +
 .../qwen3-coder-480b/r512/run_metadata.json   |   5 +
 .../qwen3-coder-480b/r512/src/vllm/.gitkeep   |   0
 .../r64/accuracy/accuracy.txt                 |   1 +
 .../r64/accuracy/accuracy_result.json         |   6 +
 .../r64}/mlperf_endpoints_log_detail.json     |   0
 .../r64}/mlperf_endpoints_log_summary.json    |   0
 .../qwen3-coder-480b/r64/point.yaml}          |   0
 .../qwen3-coder-480b/r64/report.txt           |   1 +
 .../qwen3-coder-480b/r64/run_metadata.json    |   5 +
 .../qwen3-coder-480b/r64/src/vllm/.gitkeep    |   0
 .../qwen3-coder-480b/r8/accuracy/accuracy.txt |   1 +
 .../r8/accuracy/accuracy_result.json          |   6 +
 .../r8}/mlperf_endpoints_log_detail.json      |   0
 .../r8}/mlperf_endpoints_log_summary.json     |   0
 .../qwen3-coder-480b/r8/point.yaml}           |   0
 .../qwen3-coder-480b/r8/report.txt            |   1 +
 .../qwen3-coder-480b/r8/run_metadata.json     |   5 +
 .../qwen3-coder-480b/r8/src/vllm/.gitkeep     |   0
 .../qwen3-coder-480b/sweep_distributions.csv  |   1 +
 .../qwen3-coder-480b/sweep_summary.csv        |   1 +
 .../system_desc.json}                         |   0
 .../r1024/accuracy/accuracy.txt               |   1 +
 .../r1024/accuracy/accuracy_result.json       |   6 +
 .../r1024}/mlperf_endpoints_log_detail.json   |   0
 .../r1024}/mlperf_endpoints_log_summary.json  |   0
 .../qwen3-coder-480b/r1024/point.yaml}        |   0
 .../qwen3-coder-480b/r1024/report.txt         |   1 +
 .../qwen3-coder-480b/r1024/run_metadata.json  |   5 +
 .../qwen3-coder-480b/r1024/src/vllm/.gitkeep  |   0
 .../r128/accuracy/accuracy.txt                |   1 +
 .../r128/accuracy/accuracy_result.json        |   6 +
 .../r128}/mlperf_endpoints_log_detail.json    |   0
 .../r128}/mlperf_endpoints_log_summary.json   |   0
 .../qwen3-coder-480b/r128/point.yaml}         |   0
 .../qwen3-coder-480b/r128/report.txt          |   1 +
 .../qwen3-coder-480b/r128/run_metadata.json   |   5 +
 .../qwen3-coder-480b/r128/src/vllm/.gitkeep   |   0
 .../r16/accuracy/accuracy.txt                 |   1 +
 .../r16/accuracy/accuracy_result.json         |   6 +
 .../r16}/mlperf_endpoints_log_detail.json     |   0
 .../r16}/mlperf_endpoints_log_summary.json    |   0
 .../qwen3-coder-480b/r16/point.yaml}          |   0
 .../qwen3-coder-480b/r16/report.txt           |   1 +
 .../qwen3-coder-480b/r16/run_metadata.json    |   5 +
 .../qwen3-coder-480b/r16/src/vllm/.gitkeep    |   0
 .../r256/accuracy/accuracy.txt                |   1 +
 .../r256/accuracy/accuracy_result.json        |   6 +
 .../r256}/mlperf_endpoints_log_detail.json    |   0
 .../r256}/mlperf_endpoints_log_summary.json   |   0
 .../qwen3-coder-480b/r256/point.yaml}         |   0
 .../qwen3-coder-480b/r256/report.txt          |   1 +
 .../qwen3-coder-480b/r256/run_metadata.json   |   5 +
 .../qwen3-coder-480b/r256/src/vllm/.gitkeep   |   0
 .../r32/accuracy/accuracy.txt                 |   1 +
 .../r32/accuracy/accuracy_result.json         |   6 +
 .../r32}/mlperf_endpoints_log_detail.json     |   0
 .../r32}/mlperf_endpoints_log_summary.json    |   0
 .../qwen3-coder-480b/r32/point.yaml}          |   0
 .../qwen3-coder-480b/r32/report.txt           |   1 +
 .../qwen3-coder-480b/r32/run_metadata.json    |   5 +
 .../qwen3-coder-480b/r32/src/vllm/.gitkeep    |   0
 .../r512/accuracy/accuracy.txt                |   1 +
 .../r512/accuracy/accuracy_result.json        |   6 +
 .../r512}/mlperf_endpoints_log_detail.json    |   0
 .../r512}/mlperf_endpoints_log_summary.json   |   0
 .../qwen3-coder-480b/r512/point.yaml}         |   0
 .../qwen3-coder-480b/r512/report.txt          |   1 +
 .../qwen3-coder-480b/r512/run_metadata.json   |   5 +
 .../qwen3-coder-480b/r512/src/vllm/.gitkeep   |   0
 .../r64/accuracy/accuracy.txt                 |   1 +
 .../r64/accuracy/accuracy_result.json         |   6 +
 .../r64}/mlperf_endpoints_log_detail.json     |   0
 .../r64}/mlperf_endpoints_log_summary.json    |   0
 .../qwen3-coder-480b/r64/point.yaml}          |   0
 .../qwen3-coder-480b/r64/report.txt           |   1 +
 .../qwen3-coder-480b/r64/run_metadata.json    |   5 +
 .../qwen3-coder-480b/r64/src/vllm/.gitkeep    |   0
 .../qwen3-coder-480b/r8/accuracy/accuracy.txt |   1 +
 .../r8/accuracy/accuracy_result.json          |   6 +
 .../r8}/mlperf_endpoints_log_detail.json      |   0
 .../r8}/mlperf_endpoints_log_summary.json     |   0
 .../qwen3-coder-480b/r8/point.yaml}           |   0
 .../qwen3-coder-480b/r8/report.txt            |   1 +
 .../qwen3-coder-480b/r8/run_metadata.json     |   5 +
 .../qwen3-coder-480b/r8/src/vllm/.gitkeep     |   0
 .../qwen3-coder-480b/sweep_distributions.csv  |   1 +
 .../qwen3-coder-480b/sweep_summary.csv        |   1 +
 .../system_desc.json}                         |   0
 .../llama3-8b/r1/accuracy/accuracy.txt        |   1 +
 .../r1/accuracy/accuracy_result.json          |   6 +
 .../r1}/mlperf_endpoints_log_detail.json      |   0
 .../r1}/mlperf_endpoints_log_summary.json     |   0
 .../llama3-8b/r1/point.yaml}                  |   0
 .../sys_gaudi_dp1/llama3-8b/r1/report.txt     |   1 +
 .../llama3-8b/r1/run_metadata.json            |   5 +
 .../llama3-8b/r1/src/vllm/.gitkeep            |   0
 .../llama3-8b/r1024/accuracy/accuracy.txt     |   1 +
 .../r1024/accuracy/accuracy_result.json       |   6 +
 .../r1024}/mlperf_endpoints_log_detail.json   |   0
 .../r1024}/mlperf_endpoints_log_summary.json  |   0
 .../llama3-8b/r1024/point.yaml}               |   0
 .../sys_gaudi_dp1/llama3-8b/r1024/report.txt  |   1 +
 .../llama3-8b/r1024/run_metadata.json         |   5 +
 .../llama3-8b/r1024/src/vllm/.gitkeep         |   0
 .../llama3-8b/r128/accuracy/accuracy.txt      |   1 +
 .../r128/accuracy/accuracy_result.json        |   6 +
 .../r128}/mlperf_endpoints_log_detail.json    |   0
 .../r128}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-8b/r128/point.yaml}                |   0
 .../sys_gaudi_dp1/llama3-8b/r128/report.txt   |   1 +
 .../llama3-8b/r128/run_metadata.json          |   5 +
 .../llama3-8b/r128/src/vllm/.gitkeep          |   0
 .../llama3-8b/r16/accuracy/accuracy.txt       |   1 +
 .../r16/accuracy/accuracy_result.json         |   6 +
 .../r16}/mlperf_endpoints_log_detail.json     |   0
 .../r16}/mlperf_endpoints_log_summary.json    |   0
 .../llama3-8b/r16/point.yaml}                 |   0
 .../sys_gaudi_dp1/llama3-8b/r16/report.txt    |   1 +
 .../llama3-8b/r16/run_metadata.json           |   5 +
 .../llama3-8b/r16/src/vllm/.gitkeep           |   0
 .../llama3-8b/r2/accuracy/accuracy.txt        |   1 +
 .../r2/accuracy/accuracy_result.json          |   6 +
 .../r2}/mlperf_endpoints_log_detail.json      |   0
 .../r2}/mlperf_endpoints_log_summary.json     |   0
 .../llama3-8b/r2/point.yaml}                  |   0
 .../sys_gaudi_dp1/llama3-8b/r2/report.txt     |   1 +
 .../llama3-8b/r2/run_metadata.json            |   5 +
 .../llama3-8b/r2/src/vllm/.gitkeep            |   0
 .../llama3-8b/r256/accuracy/accuracy.txt      |   1 +
 .../r256/accuracy/accuracy_result.json        |   6 +
 .../r256}/mlperf_endpoints_log_detail.json    |   0
 .../r256}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-8b/r256/point.yaml}                |   0
 .../sys_gaudi_dp1/llama3-8b/r256/report.txt   |   1 +
 .../llama3-8b/r256/run_metadata.json          |   5 +
 .../llama3-8b/r256/src/vllm/.gitkeep          |   0
 .../llama3-8b/r32/accuracy/accuracy.txt       |   1 +
 .../r32/accuracy/accuracy_result.json         |   6 +
 .../r32}/mlperf_endpoints_log_detail.json     |   0
 .../r32}/mlperf_endpoints_log_summary.json    |   0
 .../llama3-8b/r32/point.yaml}                 |   0
 .../sys_gaudi_dp1/llama3-8b/r32/report.txt    |   1 +
 .../llama3-8b/r32/run_metadata.json           |   5 +
 .../llama3-8b/r32/src/vllm/.gitkeep           |   0
 .../llama3-8b/r4/accuracy/accuracy.txt        |   1 +
 .../r4/accuracy/accuracy_result.json          |   6 +
 .../r4}/mlperf_endpoints_log_detail.json      |   0
 .../r4}/mlperf_endpoints_log_summary.json     |   0
 .../llama3-8b/r4/point.yaml}                  |   0
 .../sys_gaudi_dp1/llama3-8b/r4/report.txt     |   1 +
 .../llama3-8b/r4/run_metadata.json            |   5 +
 .../llama3-8b/r4/src/vllm/.gitkeep            |   0
 .../llama3-8b/r512/accuracy/accuracy.txt      |   1 +
 .../r512/accuracy/accuracy_result.json        |   6 +
 .../r512}/mlperf_endpoints_log_detail.json    |   0
 .../r512}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-8b/r512/point.yaml}                |   0
 .../sys_gaudi_dp1/llama3-8b/r512/report.txt   |   1 +
 .../llama3-8b/r512/run_metadata.json          |   5 +
 .../llama3-8b/r512/src/vllm/.gitkeep          |   0
 .../llama3-8b/r64/accuracy/accuracy.txt       |   1 +
 .../r64/accuracy/accuracy_result.json         |   6 +
 .../r64}/mlperf_endpoints_log_detail.json     |   0
 .../r64}/mlperf_endpoints_log_summary.json    |   0
 .../llama3-8b/r64/point.yaml}                 |   0
 .../sys_gaudi_dp1/llama3-8b/r64/report.txt    |   1 +
 .../llama3-8b/r64/run_metadata.json           |   5 +
 .../llama3-8b/r64/src/vllm/.gitkeep           |   0
 .../llama3-8b/r8/accuracy/accuracy.txt        |   1 +
 .../r8/accuracy/accuracy_result.json          |   6 +
 .../r8}/mlperf_endpoints_log_detail.json      |   0
 .../r8}/mlperf_endpoints_log_summary.json     |   0
 .../llama3-8b/r8/point.yaml}                  |   0
 .../sys_gaudi_dp1/llama3-8b/r8/report.txt     |   1 +
 .../llama3-8b/r8/run_metadata.json            |   5 +
 .../llama3-8b/r8/src/vllm/.gitkeep            |   0
 .../llama3-8b/sweep_distributions.csv         |   1 +
 .../sys_gaudi_dp1/llama3-8b/sweep_summary.csv |   1 +
 .../system_desc.json}                         |   0
 .../llama3-8b/r1/accuracy/accuracy.txt        |   1 +
 .../r1/accuracy/accuracy_result.json          |   6 +
 .../r1}/mlperf_endpoints_log_detail.json      |   0
 .../r1}/mlperf_endpoints_log_summary.json     |   0
 .../llama3-8b/r1/point.yaml}                  |   0
 .../sys_gaudi_dp2/llama3-8b/r1/report.txt     |   1 +
 .../llama3-8b/r1/run_metadata.json            |   5 +
 .../llama3-8b/r1/src/vllm/.gitkeep            |   0
 .../llama3-8b/r1024/accuracy/accuracy.txt     |   1 +
 .../r1024/accuracy/accuracy_result.json       |   6 +
 .../r1024}/mlperf_endpoints_log_detail.json   |   0
 .../r1024}/mlperf_endpoints_log_summary.json  |   0
 .../llama3-8b/r1024/point.yaml}               |   0
 .../sys_gaudi_dp2/llama3-8b/r1024/report.txt  |   1 +
 .../llama3-8b/r1024/run_metadata.json         |   5 +
 .../llama3-8b/r1024/src/vllm/.gitkeep         |   0
 .../llama3-8b/r128/accuracy/accuracy.txt      |   1 +
 .../r128/accuracy/accuracy_result.json        |   6 +
 .../r128}/mlperf_endpoints_log_detail.json    |   0
 .../r128}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-8b/r128/point.yaml}                |   0
 .../sys_gaudi_dp2/llama3-8b/r128/report.txt   |   1 +
 .../llama3-8b/r128/run_metadata.json          |   5 +
 .../llama3-8b/r128/src/vllm/.gitkeep          |   0
 .../llama3-8b/r16/accuracy/accuracy.txt       |   1 +
 .../r16/accuracy/accuracy_result.json         |   6 +
 .../r16}/mlperf_endpoints_log_detail.json     |   0
 .../r16}/mlperf_endpoints_log_summary.json    |   0
 .../llama3-8b/r16/point.yaml}                 |   0
 .../sys_gaudi_dp2/llama3-8b/r16/report.txt    |   1 +
 .../llama3-8b/r16/run_metadata.json           |   5 +
 .../llama3-8b/r16/src/vllm/.gitkeep           |   0
 .../llama3-8b/r2/accuracy/accuracy.txt        |   1 +
 .../r2/accuracy/accuracy_result.json          |   6 +
 .../r2}/mlperf_endpoints_log_detail.json      |   0
 .../r2}/mlperf_endpoints_log_summary.json     |   0
 .../llama3-8b/r2/point.yaml}                  |   0
 .../sys_gaudi_dp2/llama3-8b/r2/report.txt     |   1 +
 .../llama3-8b/r2/run_metadata.json            |   5 +
 .../llama3-8b/r2/src/vllm/.gitkeep            |   0
 .../llama3-8b/r256/accuracy/accuracy.txt      |   1 +
 .../r256/accuracy/accuracy_result.json        |   6 +
 .../r256}/mlperf_endpoints_log_detail.json    |   0
 .../r256}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-8b/r256/point.yaml}                |   0
 .../sys_gaudi_dp2/llama3-8b/r256/report.txt   |   1 +
 .../llama3-8b/r256/run_metadata.json          |   5 +
 .../llama3-8b/r256/src/vllm/.gitkeep          |   0
 .../llama3-8b/r32/accuracy/accuracy.txt       |   1 +
 .../r32/accuracy/accuracy_result.json         |   6 +
 .../r32}/mlperf_endpoints_log_detail.json     |   0
 .../r32}/mlperf_endpoints_log_summary.json    |   0
 .../llama3-8b/r32/point.yaml}                 |   0
 .../sys_gaudi_dp2/llama3-8b/r32/report.txt    |   1 +
 .../llama3-8b/r32/run_metadata.json           |   5 +
 .../llama3-8b/r32/src/vllm/.gitkeep           |   0
 .../llama3-8b/r4/accuracy/accuracy.txt        |   1 +
 .../r4/accuracy/accuracy_result.json          |   6 +
 .../r4}/mlperf_endpoints_log_detail.json      |   0
 .../r4}/mlperf_endpoints_log_summary.json     |   0
 .../llama3-8b/r4/point.yaml}                  |   0
 .../sys_gaudi_dp2/llama3-8b/r4/report.txt     |   1 +
 .../llama3-8b/r4/run_metadata.json            |   5 +
 .../llama3-8b/r4/src/vllm/.gitkeep            |   0
 .../llama3-8b/r512/accuracy/accuracy.txt      |   1 +
 .../r512/accuracy/accuracy_result.json        |   6 +
 .../r512}/mlperf_endpoints_log_detail.json    |   0
 .../r512}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-8b/r512/point.yaml}                |   0
 .../sys_gaudi_dp2/llama3-8b/r512/report.txt   |   1 +
 .../llama3-8b/r512/run_metadata.json          |   5 +
 .../llama3-8b/r512/src/vllm/.gitkeep          |   0
 .../llama3-8b/r64/accuracy/accuracy.txt       |   1 +
 .../r64/accuracy/accuracy_result.json         |   6 +
 .../r64}/mlperf_endpoints_log_detail.json     |   0
 .../r64}/mlperf_endpoints_log_summary.json    |   0
 .../llama3-8b/r64/point.yaml}                 |   0
 .../sys_gaudi_dp2/llama3-8b/r64/report.txt    |   1 +
 .../llama3-8b/r64/run_metadata.json           |   5 +
 .../llama3-8b/r64/src/vllm/.gitkeep           |   0
 .../llama3-8b/r8/accuracy/accuracy.txt        |   1 +
 .../r8/accuracy/accuracy_result.json          |   6 +
 .../r8}/mlperf_endpoints_log_detail.json      |   0
 .../r8}/mlperf_endpoints_log_summary.json     |   0
 .../llama3-8b/r8/point.yaml}                  |   0
 .../sys_gaudi_dp2/llama3-8b/r8/report.txt     |   1 +
 .../llama3-8b/r8/run_metadata.json            |   5 +
 .../llama3-8b/r8/src/vllm/.gitkeep            |   0
 .../llama3-8b/sweep_distributions.csv         |   1 +
 .../sys_gaudi_dp2/llama3-8b/sweep_summary.csv |   1 +
 .../system_desc.json}                         |   0
 .../llama3-70b/r1024/accuracy/accuracy.txt    |   1 +
 .../r1024/accuracy/accuracy_result.json       |   6 +
 .../r1024}/mlperf_endpoints_log_detail.json   |   0
 .../r1024}/mlperf_endpoints_log_summary.json  |   0
 .../llama3-70b/r1024/point.yaml}              |   0
 .../sys_8gpu_vllm/llama3-70b/r1024/report.txt |   1 +
 .../llama3-70b/r1024/run_metadata.json        |   5 +
 .../llama3-70b/r1024/src/vllm/.gitkeep        |   0
 .../llama3-70b/r128/accuracy/accuracy.txt     |   1 +
 .../r128/accuracy/accuracy_result.json        |   6 +
 .../r128}/mlperf_endpoints_log_detail.json    |   0
 .../r128}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-70b/r128/point.yaml}               |   0
 .../sys_8gpu_vllm/llama3-70b/r128/report.txt  |   1 +
 .../llama3-70b/r128/run_metadata.json         |   5 +
 .../llama3-70b/r128/src/vllm/.gitkeep         |   0
 .../llama3-70b/r1536/accuracy/accuracy.txt    |   1 +
 .../r1536/accuracy/accuracy_result.json       |   6 +
 .../r1536}/mlperf_endpoints_log_detail.json   |   0
 .../r1536}/mlperf_endpoints_log_summary.json  |   0
 .../llama3-70b/r1536/point.yaml}              |   0
 .../sys_8gpu_vllm/llama3-70b/r1536/report.txt |   1 +
 .../llama3-70b/r1536/run_metadata.json        |   5 +
 .../llama3-70b/r1536/src/vllm/.gitkeep        |   0
 .../llama3-70b/r192/accuracy/accuracy.txt     |   1 +
 .../r192/accuracy/accuracy_result.json        |   6 +
 .../r192}/mlperf_endpoints_log_detail.json    |   0
 .../r192}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-70b/r192/point.yaml}               |   0
 .../sys_8gpu_vllm/llama3-70b/r192/report.txt  |   1 +
 .../llama3-70b/r192/run_metadata.json         |   5 +
 .../llama3-70b/r192/src/vllm/.gitkeep         |   0
 .../llama3-70b/r2048/accuracy/accuracy.txt    |   1 +
 .../r2048/accuracy/accuracy_result.json       |   6 +
 .../r2048}/mlperf_endpoints_log_detail.json   |   0
 .../r2048}/mlperf_endpoints_log_summary.json  |   0
 .../llama3-70b/r2048/point.yaml}              |   0
 .../sys_8gpu_vllm/llama3-70b/r2048/report.txt |   1 +
 .../llama3-70b/r2048/run_metadata.json        |   5 +
 .../llama3-70b/r2048/src/vllm/.gitkeep        |   0
 .../llama3-70b/r256/accuracy/accuracy.txt     |   1 +
 .../r256/accuracy/accuracy_result.json        |   6 +
 .../r256}/mlperf_endpoints_log_detail.json    |   0
 .../r256}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-70b/r256/point.yaml}               |   0
 .../sys_8gpu_vllm/llama3-70b/r256/report.txt  |   1 +
 .../llama3-70b/r256/run_metadata.json         |   5 +
 .../llama3-70b/r256/src/vllm/.gitkeep         |   0
 .../llama3-70b/r384/accuracy/accuracy.txt     |   1 +
 .../r384/accuracy/accuracy_result.json        |   6 +
 .../r384}/mlperf_endpoints_log_detail.json    |   0
 .../r384}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-70b/r384/point.yaml}               |   0
 .../sys_8gpu_vllm/llama3-70b/r384/report.txt  |   1 +
 .../llama3-70b/r384/run_metadata.json         |   5 +
 .../llama3-70b/r384/src/vllm/.gitkeep         |   0
 .../llama3-70b/r512/accuracy/accuracy.txt     |   1 +
 .../r512/accuracy/accuracy_result.json        |   6 +
 .../r512}/mlperf_endpoints_log_detail.json    |   0
 .../r512}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-70b/r512/point.yaml}               |   0
 .../sys_8gpu_vllm/llama3-70b/r512/report.txt  |   1 +
 .../llama3-70b/r512/run_metadata.json         |   5 +
 .../llama3-70b/r512/src/vllm/.gitkeep         |   0
 .../llama3-70b/r64/accuracy/accuracy.txt      |   1 +
 .../r64/accuracy/accuracy_result.json         |   6 +
 .../r64}/mlperf_endpoints_log_detail.json     |   0
 .../r64}/mlperf_endpoints_log_summary.json    |   0
 .../llama3-70b/r64/point.yaml}                |   0
 .../sys_8gpu_vllm/llama3-70b/r64/report.txt   |   1 +
 .../llama3-70b/r64/run_metadata.json          |   5 +
 .../llama3-70b/r64/src/vllm/.gitkeep          |   0
 .../llama3-70b/r768/accuracy/accuracy.txt     |   1 +
 .../r768/accuracy/accuracy_result.json        |   6 +
 .../r768}/mlperf_endpoints_log_detail.json    |   0
 .../r768}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-70b/r768/point.yaml}               |   0
 .../sys_8gpu_vllm/llama3-70b/r768/report.txt  |   1 +
 .../llama3-70b/r768/run_metadata.json         |   5 +
 .../llama3-70b/r768/src/vllm/.gitkeep         |   0
 .../llama3-70b/sweep_distributions.csv        |   1 +
 .../llama3-70b/sweep_summary.csv              |   1 +
 .../system_desc.json}                         |   0
 .../llama3-70b/r1024/accuracy/accuracy.txt    |   1 +
 .../r1024/accuracy/accuracy_result.json       |   6 +
 .../r1024}/mlperf_endpoints_log_detail.json   |   0
 .../r1024}/mlperf_endpoints_log_summary.json  |   0
 .../llama3-70b/r1024/point.yaml}              |   0
 .../llama3-70b/r1024/report.txt               |   1 +
 .../llama3-70b/r1024/run_metadata.json        |   5 +
 .../llama3-70b/r1024/src/vllm/.gitkeep        |   0
 .../llama3-70b/r128/accuracy/accuracy.txt     |   1 +
 .../r128/accuracy/accuracy_result.json        |   6 +
 .../r128}/mlperf_endpoints_log_detail.json    |   0
 .../r128}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-70b/r128/point.yaml}               |   0
 .../llama3-70b/r128/report.txt                |   1 +
 .../llama3-70b/r128/run_metadata.json         |   5 +
 .../llama3-70b/r128/src/vllm/.gitkeep         |   0
 .../llama3-70b/r1536/accuracy/accuracy.txt    |   1 +
 .../r1536/accuracy/accuracy_result.json       |   6 +
 .../r1536}/mlperf_endpoints_log_detail.json   |   0
 .../r1536}/mlperf_endpoints_log_summary.json  |   0
 .../llama3-70b/r1536/point.yaml}              |   0
 .../llama3-70b/r1536/report.txt               |   1 +
 .../llama3-70b/r1536/run_metadata.json        |   5 +
 .../llama3-70b/r1536/src/vllm/.gitkeep        |   0
 .../llama3-70b/r192/accuracy/accuracy.txt     |   1 +
 .../r192/accuracy/accuracy_result.json        |   6 +
 .../r192}/mlperf_endpoints_log_detail.json    |   0
 .../r192}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-70b/r192/point.yaml}               |   0
 .../llama3-70b/r192/report.txt                |   1 +
 .../llama3-70b/r192/run_metadata.json         |   5 +
 .../llama3-70b/r192/src/vllm/.gitkeep         |   0
 .../llama3-70b/r2048/accuracy/accuracy.txt    |   1 +
 .../r2048/accuracy/accuracy_result.json       |   6 +
 .../r2048}/mlperf_endpoints_log_detail.json   |   0
 .../r2048}/mlperf_endpoints_log_summary.json  |   0
 .../llama3-70b/r2048/point.yaml}              |   0
 .../llama3-70b/r2048/report.txt               |   1 +
 .../llama3-70b/r2048/run_metadata.json        |   5 +
 .../llama3-70b/r2048/src/vllm/.gitkeep        |   0
 .../llama3-70b/r256/accuracy/accuracy.txt     |   1 +
 .../r256/accuracy/accuracy_result.json        |   6 +
 .../r256}/mlperf_endpoints_log_detail.json    |   0
 .../r256}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-70b/r256/point.yaml}               |   0
 .../llama3-70b/r256/report.txt                |   1 +
 .../llama3-70b/r256/run_metadata.json         |   5 +
 .../llama3-70b/r256/src/vllm/.gitkeep         |   0
 .../llama3-70b/r384/accuracy/accuracy.txt     |   1 +
 .../r384/accuracy/accuracy_result.json        |   6 +
 .../r384}/mlperf_endpoints_log_detail.json    |   0
 .../r384}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-70b/r384/point.yaml}               |   0
 .../llama3-70b/r384/report.txt                |   1 +
 .../llama3-70b/r384/run_metadata.json         |   5 +
 .../llama3-70b/r384/src/vllm/.gitkeep         |   0
 .../llama3-70b/r512/accuracy/accuracy.txt     |   1 +
 .../r512/accuracy/accuracy_result.json        |   6 +
 .../r512}/mlperf_endpoints_log_detail.json    |   0
 .../r512}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-70b/r512/point.yaml}               |   0
 .../llama3-70b/r512/report.txt                |   1 +
 .../llama3-70b/r512/run_metadata.json         |   5 +
 .../llama3-70b/r512/src/vllm/.gitkeep         |   0
 .../llama3-70b/r64/accuracy/accuracy.txt      |   1 +
 .../r64/accuracy/accuracy_result.json         |   6 +
 .../r64}/mlperf_endpoints_log_detail.json     |   0
 .../r64}/mlperf_endpoints_log_summary.json    |   0
 .../llama3-70b/r64/point.yaml}                |   0
 .../sys_8gpu_sglang/llama3-70b/r64/report.txt |   1 +
 .../llama3-70b/r64/run_metadata.json          |   5 +
 .../llama3-70b/r64/src/vllm/.gitkeep          |   0
 .../llama3-70b/r768/accuracy/accuracy.txt     |   1 +
 .../r768/accuracy/accuracy_result.json        |   6 +
 .../r768}/mlperf_endpoints_log_detail.json    |   0
 .../r768}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-70b/r768/point.yaml}               |   0
 .../llama3-70b/r768/report.txt                |   1 +
 .../llama3-70b/r768/run_metadata.json         |   5 +
 .../llama3-70b/r768/src/vllm/.gitkeep         |   0
 .../llama3-70b/sweep_distributions.csv        |   1 +
 .../llama3-70b/sweep_summary.csv              |   1 +
 .../system_desc.json}                         |   0
 .../deepseek-r1/r1/accuracy/accuracy.txt      |   1 +
 .../r1/accuracy/accuracy_result.json          |   6 +
 .../r1}/mlperf_endpoints_log_detail.json      |   0
 .../r1}/mlperf_endpoints_log_summary.json     |   0
 .../deepseek-r1/r1/point.yaml}                |   0
 .../sys_h200_8gpu/deepseek-r1/r1/report.txt   |   1 +
 .../deepseek-r1/r1/run_metadata.json          |   5 +
 .../deepseek-r1/r1/src/vllm/.gitkeep          |   0
 .../deepseek-r1/r128/accuracy/accuracy.txt    |   1 +
 .../r128/accuracy/accuracy_result.json        |   6 +
 .../r128}/mlperf_endpoints_log_detail.json    |   0
 .../r128}/mlperf_endpoints_log_summary.json   |   0
 .../deepseek-r1/r128/point.yaml}              |   0
 .../sys_h200_8gpu/deepseek-r1/r128/report.txt |   1 +
 .../deepseek-r1/r128/run_metadata.json        |   5 +
 .../deepseek-r1/r128/src/vllm/.gitkeep        |   0
 .../deepseek-r1/r16/accuracy/accuracy.txt     |   1 +
 .../r16/accuracy/accuracy_result.json         |   6 +
 .../r16}/mlperf_endpoints_log_detail.json     |   0
 .../r16}/mlperf_endpoints_log_summary.json    |   0
 .../deepseek-r1/r16/point.yaml}               |   0
 .../sys_h200_8gpu/deepseek-r1/r16/report.txt  |   1 +
 .../deepseek-r1/r16/run_metadata.json         |   5 +
 .../deepseek-r1/r16/src/vllm/.gitkeep         |   0
 .../deepseek-r1/r2/accuracy/accuracy.txt      |   1 +
 .../r2/accuracy/accuracy_result.json          |   6 +
 .../r2}/mlperf_endpoints_log_detail.json      |   0
 .../r2}/mlperf_endpoints_log_summary.json     |   0
 .../deepseek-r1/r2/point.yaml}                |   0
 .../sys_h200_8gpu/deepseek-r1/r2/report.txt   |   1 +
 .../deepseek-r1/r2/run_metadata.json          |   5 +
 .../deepseek-r1/r2/src/vllm/.gitkeep          |   0
 .../deepseek-r1/r256/accuracy/accuracy.txt    |   1 +
 .../r256/accuracy/accuracy_result.json        |   6 +
 .../r256}/mlperf_endpoints_log_detail.json    |   0
 .../r256}/mlperf_endpoints_log_summary.json   |   0
 .../deepseek-r1/r256/point.yaml}              |   0
 .../sys_h200_8gpu/deepseek-r1/r256/report.txt |   1 +
 .../deepseek-r1/r256/run_metadata.json        |   5 +
 .../deepseek-r1/r256/src/vllm/.gitkeep        |   0
 .../deepseek-r1/r32/accuracy/accuracy.txt     |   1 +
 .../r32/accuracy/accuracy_result.json         |   6 +
 .../r32}/mlperf_endpoints_log_detail.json     |   0
 .../r32}/mlperf_endpoints_log_summary.json    |   0
 .../deepseek-r1/r32/point.yaml}               |   0
 .../sys_h200_8gpu/deepseek-r1/r32/report.txt  |   1 +
 .../deepseek-r1/r32/run_metadata.json         |   5 +
 .../deepseek-r1/r32/src/vllm/.gitkeep         |   0
 .../deepseek-r1/r4/accuracy/accuracy.txt      |   1 +
 .../r4/accuracy/accuracy_result.json          |   6 +
 .../r4}/mlperf_endpoints_log_detail.json      |   0
 .../r4}/mlperf_endpoints_log_summary.json     |   0
 .../deepseek-r1/r4/point.yaml}                |   0
 .../sys_h200_8gpu/deepseek-r1/r4/report.txt   |   1 +
 .../deepseek-r1/r4/run_metadata.json          |   5 +
 .../deepseek-r1/r4/src/vllm/.gitkeep          |   0
 .../deepseek-r1/r512/accuracy/accuracy.txt    |   1 +
 .../r512/accuracy/accuracy_result.json        |   6 +
 .../r512}/mlperf_endpoints_log_detail.json    |   0
 .../r512}/mlperf_endpoints_log_summary.json   |   0
 .../deepseek-r1/r512/point.yaml}              |   0
 .../sys_h200_8gpu/deepseek-r1/r512/report.txt |   1 +
 .../deepseek-r1/r512/run_metadata.json        |   5 +
 .../deepseek-r1/r512/src/vllm/.gitkeep        |   0
 .../deepseek-r1/r64/accuracy/accuracy.txt     |   1 +
 .../r64/accuracy/accuracy_result.json         |   6 +
 .../r64}/mlperf_endpoints_log_detail.json     |   0
 .../r64}/mlperf_endpoints_log_summary.json    |   0
 .../deepseek-r1/r64/point.yaml}               |   0
 .../sys_h200_8gpu/deepseek-r1/r64/report.txt  |   1 +
 .../deepseek-r1/r64/run_metadata.json         |   5 +
 .../deepseek-r1/r64/src/vllm/.gitkeep         |   0
 .../deepseek-r1/r8/accuracy/accuracy.txt      |   1 +
 .../r8/accuracy/accuracy_result.json          |   6 +
 .../r8}/mlperf_endpoints_log_detail.json      |   0
 .../r8}/mlperf_endpoints_log_summary.json     |   0
 .../deepseek-r1/r8/point.yaml}                |   0
 .../sys_h200_8gpu/deepseek-r1/r8/report.txt   |   1 +
 .../deepseek-r1/r8/run_metadata.json          |   5 +
 .../deepseek-r1/r8/src/vllm/.gitkeep          |   0
 .../deepseek-r1/sweep_distributions.csv       |   1 +
 .../deepseek-r1/sweep_summary.csv             |   1 +
 .../system_desc.json}                         |   0
 .../deepseek-r1/r1024/accuracy/accuracy.txt   |   1 +
 .../r1024/accuracy/accuracy_result.json       |   6 +
 .../r1024}/mlperf_endpoints_log_detail.json   |   0
 .../r1024}/mlperf_endpoints_log_summary.json  |   0
 .../deepseek-r1/r1024/point.yaml}             |   0
 .../deepseek-r1/r1024/report.txt              |   1 +
 .../deepseek-r1/r1024/run_metadata.json       |   5 +
 .../deepseek-r1/r1024/src/vllm/.gitkeep       |   0
 .../deepseek-r1/r128/accuracy/accuracy.txt    |   1 +
 .../r128/accuracy/accuracy_result.json        |   6 +
 .../r128}/mlperf_endpoints_log_detail.json    |   0
 .../r128}/mlperf_endpoints_log_summary.json   |   0
 .../deepseek-r1/r128/point.yaml}              |   0
 .../deepseek-r1/r128/report.txt               |   1 +
 .../deepseek-r1/r128/run_metadata.json        |   5 +
 .../deepseek-r1/r128/src/vllm/.gitkeep        |   0
 .../deepseek-r1/r16384/accuracy/accuracy.txt  |   1 +
 .../r16384/accuracy/accuracy_result.json      |   6 +
 .../r16384}/mlperf_endpoints_log_detail.json  |   0
 .../r16384}/mlperf_endpoints_log_summary.json |   0
 .../deepseek-r1/r16384/point.yaml}            |   0
 .../deepseek-r1/r16384/report.txt             |   1 +
 .../deepseek-r1/r16384/run_metadata.json      |   5 +
 .../deepseek-r1/r16384/src/vllm/.gitkeep      |   0
 .../deepseek-r1/r2048/accuracy/accuracy.txt   |   1 +
 .../r2048/accuracy/accuracy_result.json       |   6 +
 .../r2048}/mlperf_endpoints_log_detail.json   |   0
 .../r2048}/mlperf_endpoints_log_summary.json  |   0
 .../deepseek-r1/r2048/point.yaml}             |   0
 .../deepseek-r1/r2048/report.txt              |   1 +
 .../deepseek-r1/r2048/run_metadata.json       |   5 +
 .../deepseek-r1/r2048/src/vllm/.gitkeep       |   0
 .../deepseek-r1/r256/accuracy/accuracy.txt    |   1 +
 .../r256/accuracy/accuracy_result.json        |   6 +
 .../r256}/mlperf_endpoints_log_detail.json    |   0
 .../r256}/mlperf_endpoints_log_summary.json   |   0
 .../deepseek-r1/r256/point.yaml}              |   0
 .../deepseek-r1/r256/report.txt               |   1 +
 .../deepseek-r1/r256/run_metadata.json        |   5 +
 .../deepseek-r1/r256/src/vllm/.gitkeep        |   0
 .../deepseek-r1/r32/accuracy/accuracy.txt     |   1 +
 .../r32/accuracy/accuracy_result.json         |   6 +
 .../r32}/mlperf_endpoints_log_detail.json     |   0
 .../r32}/mlperf_endpoints_log_summary.json    |   0
 .../deepseek-r1/r32/point.yaml}               |   0
 .../deepseek-r1/r32/report.txt                |   1 +
 .../deepseek-r1/r32/run_metadata.json         |   5 +
 .../deepseek-r1/r32/src/vllm/.gitkeep         |   0
 .../deepseek-r1/r4096/accuracy/accuracy.txt   |   1 +
 .../r4096/accuracy/accuracy_result.json       |   6 +
 .../r4096}/mlperf_endpoints_log_detail.json   |   0
 .../r4096}/mlperf_endpoints_log_summary.json  |   0
 .../deepseek-r1/r4096/point.yaml}             |   0
 .../deepseek-r1/r4096/report.txt              |   1 +
 .../deepseek-r1/r4096/run_metadata.json       |   5 +
 .../deepseek-r1/r4096/src/vllm/.gitkeep       |   0
 .../deepseek-r1/r512/accuracy/accuracy.txt    |   1 +
 .../r512/accuracy/accuracy_result.json        |   6 +
 .../r512}/mlperf_endpoints_log_detail.json    |   0
 .../r512}/mlperf_endpoints_log_summary.json   |   0
 .../deepseek-r1/r512/point.yaml}              |   0
 .../deepseek-r1/r512/report.txt               |   1 +
 .../deepseek-r1/r512/run_metadata.json        |   5 +
 .../deepseek-r1/r512/src/vllm/.gitkeep        |   0
 .../deepseek-r1/r64/accuracy/accuracy.txt     |   1 +
 .../r64/accuracy/accuracy_result.json         |   6 +
 .../r64}/mlperf_endpoints_log_detail.json     |   0
 .../r64}/mlperf_endpoints_log_summary.json    |   0
 .../deepseek-r1/r64/point.yaml}               |   0
 .../deepseek-r1/r64/report.txt                |   1 +
 .../deepseek-r1/r64/run_metadata.json         |   5 +
 .../deepseek-r1/r64/src/vllm/.gitkeep         |   0
 .../deepseek-r1/r8192/accuracy/accuracy.txt   |   1 +
 .../r8192/accuracy/accuracy_result.json       |   6 +
 .../r8192}/mlperf_endpoints_log_detail.json   |   0
 .../r8192}/mlperf_endpoints_log_summary.json  |   0
 .../deepseek-r1/r8192/point.yaml}             |   0
 .../deepseek-r1/r8192/report.txt              |   1 +
 .../deepseek-r1/r8192/run_metadata.json       |   5 +
 .../deepseek-r1/r8192/src/vllm/.gitkeep       |   0
 .../deepseek-r1/sweep_distributions.csv       |   1 +
 .../deepseek-r1/sweep_summary.csv             |   1 +
 .../system_desc.json}                         |   0
 .../llama3-70b/r1000}/accuracy/accuracy.txt   |   0
 .../r1000}/accuracy/accuracy_result.json      |   0
 .../r1000}/mlperf_endpoints_log_detail.json   |   0
 .../r1000}/mlperf_endpoints_log_summary.json  |   0
 .../llama3-70b/r1000/point.yaml}              |   0
 .../llama3-70b/r1000/report.txt               |   1 +
 .../llama3-70b/r1000/run_metadata.json        |   5 +
 .../llama3-70b/r1000/src/vllm/.gitkeep        |   0
 .../llama3-70b/r16/accuracy/accuracy.txt      |   3 +
 .../r16/accuracy/accuracy_result.json         |   6 +
 .../r16}/mlperf_endpoints_log_detail.json     |   0
 .../r16}/mlperf_endpoints_log_summary.json    |   0
 .../llama3-70b/r16/point.yaml}                |   0
 .../acme_h100x8_001/llama3-70b/r16/report.txt |   1 +
 .../llama3-70b/r16/run_metadata.json          |   5 +
 .../llama3-70b/r16/src/vllm/.gitkeep          |   0
 .../llama3-70b/r256/accuracy/accuracy.txt     |   3 +
 .../r256/accuracy/accuracy_result.json        |   6 +
 .../r256}/mlperf_endpoints_log_detail.json    |   0
 .../r256}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-70b/r256/point.yaml}               |   0
 .../llama3-70b/r256/report.txt                |   1 +
 .../llama3-70b/r256/run_metadata.json         |   5 +
 .../llama3-70b/r256/src/vllm/.gitkeep         |   0
 .../llama3-70b/r38/accuracy/accuracy.txt      |   3 +
 .../r38/accuracy/accuracy_result.json         |   6 +
 .../r38}/mlperf_endpoints_log_detail.json     |   0
 .../r38}/mlperf_endpoints_log_summary.json    |   0
 .../llama3-70b/r38/point.yaml}                |   0
 .../acme_h100x8_001/llama3-70b/r38/report.txt |   1 +
 .../llama3-70b/r38/run_metadata.json          |   5 +
 .../llama3-70b/r38/src/vllm/.gitkeep          |   0
 .../llama3-70b/r512/accuracy/accuracy.txt     |   3 +
 .../r512/accuracy/accuracy_result.json        |   6 +
 .../r512}/mlperf_endpoints_log_detail.json    |   0
 .../r512}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-70b/r512/point.yaml}               |   0
 .../llama3-70b/r512/report.txt                |   1 +
 .../llama3-70b/r512/run_metadata.json         |   5 +
 .../llama3-70b/r512/src/vllm/.gitkeep         |   0
 .../llama3-70b/r768/accuracy/accuracy.txt     |   3 +
 .../r768/accuracy/accuracy_result.json        |   6 +
 .../r768}/mlperf_endpoints_log_detail.json    |   0
 .../r768}/mlperf_endpoints_log_summary.json   |   0
 .../llama3-70b/r768/point.yaml}               |   0
 .../llama3-70b/r768/report.txt                |   1 +
 .../llama3-70b/r768/run_metadata.json         |   5 +
 .../llama3-70b/r768/src/vllm/.gitkeep         |   0
 .../llama3-70b/r88/accuracy/accuracy.txt      |   3 +
 .../r88/accuracy/accuracy_result.json         |   6 +
 .../r88}/mlperf_endpoints_log_detail.json     |   0
 .../r88}/mlperf_endpoints_log_summary.json    |   0
 .../llama3-70b/r88/point.yaml}                |   0
 .../acme_h100x8_001/llama3-70b/r88/report.txt |   1 +
 .../llama3-70b/r88/run_metadata.json          |   5 +
 .../llama3-70b/r88/src/vllm/.gitkeep          |   0
 .../llama3-70b/sweep_distributions.csv        |   1 +
 .../llama3-70b/sweep_summary.csv              |   1 +
 .../system_desc.json}                         |   0
 .../submissions/test_builder.py               |  80 +++++--
 .../submissions/test_github.py                |   5 +-
 tests/submission_checker/conftest.py          |  10 +-
 tests/submission_checker/test_checker.py      | 215 +++++++++---------
 .../test_checks_aggregate.py                  |  25 +-
 .../test_checks_structure.py                  | 190 ++++++++++------
 856 files changed, 1872 insertions(+), 507 deletions(-)
 rename test_submissions/invalid_submission/{pareto/bad_system/llama3-70b => bad_system/llama3-70b/r16}/accuracy/accuracy.txt (100%)
 rename test_submissions/invalid_submission/{pareto/bad_system/llama3-70b => bad_system/llama3-70b/r16}/accuracy/accuracy_result.json (100%)
 rename test_submissions/invalid_submission/{pareto/bad_system/llama3-70b/results/point_16 => bad_system/llama3-70b/r16}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/invalid_submission/{pareto/bad_system/llama3-70b/results/point_16 => bad_system/llama3-70b/r16}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/invalid_submission/{pareto/bad_system/llama3-70b/points/point_16.yaml => bad_system/llama3-70b/r16/point.yaml} (100%)
 create mode 100644 test_submissions/invalid_submission/bad_system/llama3-70b/r16/report.txt
 create mode 100644 test_submissions/invalid_submission/bad_system/llama3-70b/r16/run_metadata.json
 rename test_submissions/{valid_standardized/src => invalid_submission/bad_system/llama3-70b/r16/src/vllm}/.gitkeep (100%)
 create mode 100644 test_submissions/invalid_submission/bad_system/llama3-70b/r38/accuracy/accuracy.txt
 create mode 100644 test_submissions/invalid_submission/bad_system/llama3-70b/r38/accuracy/accuracy_result.json
 rename test_submissions/invalid_submission/{pareto/bad_system/llama3-70b/results/point_38 => bad_system/llama3-70b/r38}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/invalid_submission/{pareto/bad_system/llama3-70b/results/point_38 => bad_system/llama3-70b/r38}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/invalid_submission/{pareto/bad_system/llama3-70b/points/point_38.yaml => bad_system/llama3-70b/r38/point.yaml} (100%)
 create mode 100644 test_submissions/invalid_submission/bad_system/llama3-70b/r38/report.txt
 create mode 100644 test_submissions/invalid_submission/bad_system/llama3-70b/r38/run_metadata.json
 create mode 100644 test_submissions/invalid_submission/bad_system/llama3-70b/r38/src/vllm/.gitkeep
 create mode 100644 test_submissions/invalid_submission/bad_system/llama3-70b/r88/accuracy/accuracy.txt
 create mode 100644 test_submissions/invalid_submission/bad_system/llama3-70b/r88/accuracy/accuracy_result.json
 rename test_submissions/invalid_submission/{pareto/bad_system/llama3-70b/results/point_88 => bad_system/llama3-70b/r88}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/invalid_submission/{pareto/bad_system/llama3-70b/results/point_88 => bad_system/llama3-70b/r88}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/invalid_submission/{pareto/bad_system/llama3-70b/points/point_88.yaml => bad_system/llama3-70b/r88/point.yaml} (100%)
 create mode 100644 test_submissions/invalid_submission/bad_system/llama3-70b/r88/report.txt
 create mode 100644 test_submissions/invalid_submission/bad_system/llama3-70b/r88/run_metadata.json
 create mode 100644 test_submissions/invalid_submission/bad_system/llama3-70b/r88/src/vllm/.gitkeep
 create mode 100644 test_submissions/invalid_submission/bad_system/llama3-70b/sweep_distributions.csv
 create mode 100644 test_submissions/invalid_submission/bad_system/llama3-70b/sweep_summary.csv
 rename test_submissions/invalid_submission/{systems/bad_system.json => bad_system/system_desc.json} (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b => sys_mi355x_8gpu/gpt-oss-120b/r1024}/accuracy/accuracy.txt (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b => sys_mi355x_8gpu/gpt-oss-120b/r1024}/accuracy/accuracy_result.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_1024 => sys_mi355x_8gpu/gpt-oss-120b/r1024}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_1024 => sys_mi355x_8gpu/gpt-oss-120b/r1024}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_1024.yaml => sys_mi355x_8gpu/gpt-oss-120b/r1024/point.yaml} (100%)
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/report.txt
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/run_metadata.json
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/src/vllm/.gitkeep
 rename test_submissions/{sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b => sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128}/accuracy/accuracy.txt (100%)
 rename test_submissions/{sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b => sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128}/accuracy/accuracy_result.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_128 => sys_mi355x_8gpu/gpt-oss-120b/r128}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_128 => sys_mi355x_8gpu/gpt-oss-120b/r128}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_128.yaml => sys_mi355x_8gpu/gpt-oss-120b/r128/point.yaml} (100%)
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/report.txt
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/run_metadata.json
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/src/vllm/.gitkeep
 rename test_submissions/{sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b => sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16}/accuracy/accuracy.txt (100%)
 rename test_submissions/{sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b => sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16}/accuracy/accuracy_result.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_16 => sys_mi355x_8gpu/gpt-oss-120b/r16}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_16 => sys_mi355x_8gpu/gpt-oss-120b/r16}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_16.yaml => sys_mi355x_8gpu/gpt-oss-120b/r16/point.yaml} (100%)
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/report.txt
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/run_metadata.json
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/src/vllm/.gitkeep
 rename test_submissions/{sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b => sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048}/accuracy/accuracy.txt (100%)
 rename test_submissions/{sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b => sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048}/accuracy/accuracy_result.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_2048 => sys_mi355x_8gpu/gpt-oss-120b/r2048}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_2048 => sys_mi355x_8gpu/gpt-oss-120b/r2048}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_2048.yaml => sys_mi355x_8gpu/gpt-oss-120b/r2048/point.yaml} (100%)
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/report.txt
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/run_metadata.json
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/src/vllm/.gitkeep
 rename test_submissions/{sub_e/pareto/sys_gaudi_dp1/llama3-8b => sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4}/accuracy/accuracy.txt (100%)
 rename test_submissions/{sub_e/pareto/sys_gaudi_dp1/llama3-8b => sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4}/accuracy/accuracy_result.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_4 => sys_mi355x_8gpu/gpt-oss-120b/r4}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_4 => sys_mi355x_8gpu/gpt-oss-120b/r4}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_4.yaml => sys_mi355x_8gpu/gpt-oss-120b/r4/point.yaml} (100%)
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/report.txt
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/run_metadata.json
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/src/vllm/.gitkeep
 rename test_submissions/{sub_f/pareto/sys_gaudi_dp2/llama3-8b => sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512}/accuracy/accuracy.txt (100%)
 rename test_submissions/{sub_f/pareto/sys_gaudi_dp2/llama3-8b => sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512}/accuracy/accuracy_result.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_512 => sys_mi355x_8gpu/gpt-oss-120b/r512}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_512 => sys_mi355x_8gpu/gpt-oss-120b/r512}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_512.yaml => sys_mi355x_8gpu/gpt-oss-120b/r512/point.yaml} (100%)
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/report.txt
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/run_metadata.json
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/src/vllm/.gitkeep
 rename test_submissions/{sub_g/pareto/sys_8gpu_vllm/llama3-70b => sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64}/accuracy/accuracy.txt (100%)
 rename test_submissions/{sub_g/pareto/sys_8gpu_vllm/llama3-70b => sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64}/accuracy/accuracy_result.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_64 => sys_mi355x_8gpu/gpt-oss-120b/r64}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_64 => sys_mi355x_8gpu/gpt-oss-120b/r64}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_a/{pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_64.yaml => sys_mi355x_8gpu/gpt-oss-120b/r64/point.yaml} (100%)
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/report.txt
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/run_metadata.json
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/sweep_distributions.csv
 create mode 100644 test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/sweep_summary.csv
 rename test_submissions/sub_a/{systems/sys_mi355x_8gpu.json => sys_mi355x_8gpu/system_desc.json} (100%)
 rename test_submissions/{sub_h/pareto/sys_8gpu_sglang/llama3-70b => sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024}/accuracy/accuracy.txt (100%)
 rename test_submissions/{sub_h/pareto/sys_8gpu_sglang/llama3-70b => sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024}/accuracy/accuracy_result.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_1024 => sys_mi355x_16gpu/gpt-oss-120b/r1024}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_1024 => sys_mi355x_16gpu/gpt-oss-120b/r1024}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_1024.yaml => sys_mi355x_16gpu/gpt-oss-120b/r1024/point.yaml} (100%)
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/report.txt
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/run_metadata.json
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/src/vllm/.gitkeep
 rename test_submissions/{sub_i/pareto/sys_h200_8gpu/deepseek-r1 => sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128}/accuracy/accuracy.txt (100%)
 rename test_submissions/{sub_i/pareto/sys_h200_8gpu/deepseek-r1 => sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128}/accuracy/accuracy_result.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_128 => sys_mi355x_16gpu/gpt-oss-120b/r128}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_128 => sys_mi355x_16gpu/gpt-oss-120b/r128}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_128.yaml => sys_mi355x_16gpu/gpt-oss-120b/r128/point.yaml} (100%)
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/report.txt
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/run_metadata.json
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/src/vllm/.gitkeep
 rename test_submissions/{sub_j/pareto/sys_gb300_72gpu/deepseek-r1 => sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16}/accuracy/accuracy.txt (100%)
 rename test_submissions/{sub_j/pareto/sys_gb300_72gpu/deepseek-r1 => sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16}/accuracy/accuracy_result.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_16 => sys_mi355x_16gpu/gpt-oss-120b/r16}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_16 => sys_mi355x_16gpu/gpt-oss-120b/r16}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_16.yaml => sys_mi355x_16gpu/gpt-oss-120b/r16/point.yaml} (100%)
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/report.txt
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/run_metadata.json
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/accuracy/accuracy_result.json
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_2048 => sys_mi355x_16gpu/gpt-oss-120b/r2048}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_2048 => sys_mi355x_16gpu/gpt-oss-120b/r2048}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_2048.yaml => sys_mi355x_16gpu/gpt-oss-120b/r2048/point.yaml} (100%)
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/report.txt
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/run_metadata.json
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/accuracy/accuracy_result.json
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_4 => sys_mi355x_16gpu/gpt-oss-120b/r4}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_4 => sys_mi355x_16gpu/gpt-oss-120b/r4}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_4.yaml => sys_mi355x_16gpu/gpt-oss-120b/r4/point.yaml} (100%)
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/report.txt
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/run_metadata.json
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/accuracy/accuracy_result.json
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_512 => sys_mi355x_16gpu/gpt-oss-120b/r512}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_512 => sys_mi355x_16gpu/gpt-oss-120b/r512}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_512.yaml => sys_mi355x_16gpu/gpt-oss-120b/r512/point.yaml} (100%)
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/report.txt
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/run_metadata.json
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/accuracy/accuracy_result.json
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_64 => sys_mi355x_16gpu/gpt-oss-120b/r64}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_64 => sys_mi355x_16gpu/gpt-oss-120b/r64}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_b/{pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_64.yaml => sys_mi355x_16gpu/gpt-oss-120b/r64/point.yaml} (100%)
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/report.txt
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/run_metadata.json
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/sweep_distributions.csv
 create mode 100644 test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/sweep_summary.csv
 rename test_submissions/sub_b/{systems/sys_mi355x_16gpu.json => sys_mi355x_16gpu/system_desc.json} (100%)
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/accuracy/accuracy_result.json
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_128 => sys_tpu_4chip/qwen3-coder-480b/r128}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_128 => sys_tpu_4chip/qwen3-coder-480b/r128}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_128.yaml => sys_tpu_4chip/qwen3-coder-480b/r128/point.yaml} (100%)
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/report.txt
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/run_metadata.json
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/accuracy/accuracy_result.json
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_16 => sys_tpu_4chip/qwen3-coder-480b/r16}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_16 => sys_tpu_4chip/qwen3-coder-480b/r16}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_16.yaml => sys_tpu_4chip/qwen3-coder-480b/r16/point.yaml} (100%)
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/report.txt
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/run_metadata.json
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/accuracy/accuracy_result.json
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_256 => sys_tpu_4chip/qwen3-coder-480b/r256}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_256 => sys_tpu_4chip/qwen3-coder-480b/r256}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_256.yaml => sys_tpu_4chip/qwen3-coder-480b/r256/point.yaml} (100%)
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/report.txt
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/run_metadata.json
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/accuracy/accuracy_result.json
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_32 => sys_tpu_4chip/qwen3-coder-480b/r32}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_32 => sys_tpu_4chip/qwen3-coder-480b/r32}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_32.yaml => sys_tpu_4chip/qwen3-coder-480b/r32/point.yaml} (100%)
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/report.txt
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/run_metadata.json
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/accuracy/accuracy_result.json
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_512 => sys_tpu_4chip/qwen3-coder-480b/r512}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_512 => sys_tpu_4chip/qwen3-coder-480b/r512}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_512.yaml => sys_tpu_4chip/qwen3-coder-480b/r512/point.yaml} (100%)
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/report.txt
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/run_metadata.json
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/accuracy/accuracy_result.json
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_64 => sys_tpu_4chip/qwen3-coder-480b/r64}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_64 => sys_tpu_4chip/qwen3-coder-480b/r64}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_64.yaml => sys_tpu_4chip/qwen3-coder-480b/r64/point.yaml} (100%)
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/report.txt
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/run_metadata.json
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/accuracy/accuracy_result.json
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_8 => sys_tpu_4chip/qwen3-coder-480b/r8}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_8 => sys_tpu_4chip/qwen3-coder-480b/r8}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_c/{pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_8.yaml => sys_tpu_4chip/qwen3-coder-480b/r8/point.yaml} (100%)
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/report.txt
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/run_metadata.json
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/sweep_distributions.csv
 create mode 100644 test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/sweep_summary.csv
 rename test_submissions/sub_c/{systems/sys_tpu_4chip.json => sys_tpu_4chip/system_desc.json} (100%)
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/accuracy/accuracy_result.json
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_1024 => sys_tpu_8chip/qwen3-coder-480b/r1024}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_1024 => sys_tpu_8chip/qwen3-coder-480b/r1024}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_1024.yaml => sys_tpu_8chip/qwen3-coder-480b/r1024/point.yaml} (100%)
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/report.txt
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/run_metadata.json
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/accuracy/accuracy_result.json
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_128 => sys_tpu_8chip/qwen3-coder-480b/r128}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_128 => sys_tpu_8chip/qwen3-coder-480b/r128}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_128.yaml => sys_tpu_8chip/qwen3-coder-480b/r128/point.yaml} (100%)
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/report.txt
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/run_metadata.json
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/accuracy/accuracy_result.json
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_16 => sys_tpu_8chip/qwen3-coder-480b/r16}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_16 => sys_tpu_8chip/qwen3-coder-480b/r16}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_16.yaml => sys_tpu_8chip/qwen3-coder-480b/r16/point.yaml} (100%)
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/report.txt
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/run_metadata.json
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/accuracy/accuracy_result.json
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_256 => sys_tpu_8chip/qwen3-coder-480b/r256}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_256 => sys_tpu_8chip/qwen3-coder-480b/r256}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_256.yaml => sys_tpu_8chip/qwen3-coder-480b/r256/point.yaml} (100%)
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/report.txt
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/run_metadata.json
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/accuracy/accuracy_result.json
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_32 => sys_tpu_8chip/qwen3-coder-480b/r32}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_32 => sys_tpu_8chip/qwen3-coder-480b/r32}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_32.yaml => sys_tpu_8chip/qwen3-coder-480b/r32/point.yaml} (100%)
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/report.txt
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/run_metadata.json
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/accuracy/accuracy_result.json
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_512 => sys_tpu_8chip/qwen3-coder-480b/r512}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_512 => sys_tpu_8chip/qwen3-coder-480b/r512}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_512.yaml => sys_tpu_8chip/qwen3-coder-480b/r512/point.yaml} (100%)
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/report.txt
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/run_metadata.json
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/accuracy/accuracy_result.json
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_64 => sys_tpu_8chip/qwen3-coder-480b/r64}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_64 => sys_tpu_8chip/qwen3-coder-480b/r64}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_64.yaml => sys_tpu_8chip/qwen3-coder-480b/r64/point.yaml} (100%)
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/report.txt
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/run_metadata.json
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/accuracy/accuracy_result.json
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_8 => sys_tpu_8chip/qwen3-coder-480b/r8}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_8 => sys_tpu_8chip/qwen3-coder-480b/r8}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_d/{pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_8.yaml => sys_tpu_8chip/qwen3-coder-480b/r8/point.yaml} (100%)
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/report.txt
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/run_metadata.json
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/sweep_distributions.csv
 create mode 100644 test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/sweep_summary.csv
 rename test_submissions/sub_d/{systems/sys_tpu_8chip.json => sys_tpu_8chip/system_desc.json} (100%)
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/accuracy/accuracy_result.json
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_1 => sys_gaudi_dp1/llama3-8b/r1}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_1 => sys_gaudi_dp1/llama3-8b/r1}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/points/point_1.yaml => sys_gaudi_dp1/llama3-8b/r1/point.yaml} (100%)
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/report.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/run_metadata.json
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/accuracy/accuracy_result.json
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_1024 => sys_gaudi_dp1/llama3-8b/r1024}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_1024 => sys_gaudi_dp1/llama3-8b/r1024}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/points/point_1024.yaml => sys_gaudi_dp1/llama3-8b/r1024/point.yaml} (100%)
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/report.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/run_metadata.json
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/accuracy/accuracy_result.json
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_128 => sys_gaudi_dp1/llama3-8b/r128}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_128 => sys_gaudi_dp1/llama3-8b/r128}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/points/point_128.yaml => sys_gaudi_dp1/llama3-8b/r128/point.yaml} (100%)
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/report.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/run_metadata.json
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/accuracy/accuracy_result.json
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_16 => sys_gaudi_dp1/llama3-8b/r16}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_16 => sys_gaudi_dp1/llama3-8b/r16}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/points/point_16.yaml => sys_gaudi_dp1/llama3-8b/r16/point.yaml} (100%)
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/report.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/run_metadata.json
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/accuracy/accuracy_result.json
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_2 => sys_gaudi_dp1/llama3-8b/r2}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_2 => sys_gaudi_dp1/llama3-8b/r2}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/points/point_2.yaml => sys_gaudi_dp1/llama3-8b/r2/point.yaml} (100%)
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/report.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/run_metadata.json
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/accuracy/accuracy_result.json
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_256 => sys_gaudi_dp1/llama3-8b/r256}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_256 => sys_gaudi_dp1/llama3-8b/r256}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/points/point_256.yaml => sys_gaudi_dp1/llama3-8b/r256/point.yaml} (100%)
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/report.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/run_metadata.json
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/accuracy/accuracy_result.json
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_32 => sys_gaudi_dp1/llama3-8b/r32}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_32 => sys_gaudi_dp1/llama3-8b/r32}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/points/point_32.yaml => sys_gaudi_dp1/llama3-8b/r32/point.yaml} (100%)
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/report.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/run_metadata.json
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/accuracy/accuracy_result.json
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_4 => sys_gaudi_dp1/llama3-8b/r4}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_4 => sys_gaudi_dp1/llama3-8b/r4}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/points/point_4.yaml => sys_gaudi_dp1/llama3-8b/r4/point.yaml} (100%)
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/report.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/run_metadata.json
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/accuracy/accuracy_result.json
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_512 => sys_gaudi_dp1/llama3-8b/r512}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_512 => sys_gaudi_dp1/llama3-8b/r512}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/points/point_512.yaml => sys_gaudi_dp1/llama3-8b/r512/point.yaml} (100%)
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/report.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/run_metadata.json
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/accuracy/accuracy_result.json
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_64 => sys_gaudi_dp1/llama3-8b/r64}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_64 => sys_gaudi_dp1/llama3-8b/r64}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/points/point_64.yaml => sys_gaudi_dp1/llama3-8b/r64/point.yaml} (100%)
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/report.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/run_metadata.json
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/accuracy/accuracy_result.json
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_8 => sys_gaudi_dp1/llama3-8b/r8}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/results/point_8 => sys_gaudi_dp1/llama3-8b/r8}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_e/{pareto/sys_gaudi_dp1/llama3-8b/points/point_8.yaml => sys_gaudi_dp1/llama3-8b/r8/point.yaml} (100%)
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/report.txt
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/run_metadata.json
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/sweep_distributions.csv
 create mode 100644 test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/sweep_summary.csv
 rename test_submissions/sub_e/{systems/sys_gaudi_dp1.json => sys_gaudi_dp1/system_desc.json} (100%)
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/accuracy/accuracy_result.json
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_1 => sys_gaudi_dp2/llama3-8b/r1}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_1 => sys_gaudi_dp2/llama3-8b/r1}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/points/point_1.yaml => sys_gaudi_dp2/llama3-8b/r1/point.yaml} (100%)
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/report.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/run_metadata.json
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/accuracy/accuracy_result.json
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_1024 => sys_gaudi_dp2/llama3-8b/r1024}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_1024 => sys_gaudi_dp2/llama3-8b/r1024}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/points/point_1024.yaml => sys_gaudi_dp2/llama3-8b/r1024/point.yaml} (100%)
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/report.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/run_metadata.json
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/accuracy/accuracy_result.json
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_128 => sys_gaudi_dp2/llama3-8b/r128}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_128 => sys_gaudi_dp2/llama3-8b/r128}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/points/point_128.yaml => sys_gaudi_dp2/llama3-8b/r128/point.yaml} (100%)
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/report.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/run_metadata.json
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/accuracy/accuracy_result.json
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_16 => sys_gaudi_dp2/llama3-8b/r16}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_16 => sys_gaudi_dp2/llama3-8b/r16}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/points/point_16.yaml => sys_gaudi_dp2/llama3-8b/r16/point.yaml} (100%)
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/report.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/run_metadata.json
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/accuracy/accuracy_result.json
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_2 => sys_gaudi_dp2/llama3-8b/r2}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_2 => sys_gaudi_dp2/llama3-8b/r2}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/points/point_2.yaml => sys_gaudi_dp2/llama3-8b/r2/point.yaml} (100%)
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/report.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/run_metadata.json
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/accuracy/accuracy_result.json
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_256 => sys_gaudi_dp2/llama3-8b/r256}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_256 => sys_gaudi_dp2/llama3-8b/r256}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/points/point_256.yaml => sys_gaudi_dp2/llama3-8b/r256/point.yaml} (100%)
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/report.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/run_metadata.json
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/accuracy/accuracy_result.json
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_32 => sys_gaudi_dp2/llama3-8b/r32}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_32 => sys_gaudi_dp2/llama3-8b/r32}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/points/point_32.yaml => sys_gaudi_dp2/llama3-8b/r32/point.yaml} (100%)
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/report.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/run_metadata.json
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/accuracy/accuracy_result.json
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_4 => sys_gaudi_dp2/llama3-8b/r4}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_4 => sys_gaudi_dp2/llama3-8b/r4}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/points/point_4.yaml => sys_gaudi_dp2/llama3-8b/r4/point.yaml} (100%)
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/report.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/run_metadata.json
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/accuracy/accuracy_result.json
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_512 => sys_gaudi_dp2/llama3-8b/r512}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_512 => sys_gaudi_dp2/llama3-8b/r512}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/points/point_512.yaml => sys_gaudi_dp2/llama3-8b/r512/point.yaml} (100%)
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/report.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/run_metadata.json
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/accuracy/accuracy_result.json
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_64 => sys_gaudi_dp2/llama3-8b/r64}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_64 => sys_gaudi_dp2/llama3-8b/r64}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/points/point_64.yaml => sys_gaudi_dp2/llama3-8b/r64/point.yaml} (100%)
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/report.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/run_metadata.json
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/accuracy/accuracy_result.json
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_8 => sys_gaudi_dp2/llama3-8b/r8}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/results/point_8 => sys_gaudi_dp2/llama3-8b/r8}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_f/{pareto/sys_gaudi_dp2/llama3-8b/points/point_8.yaml => sys_gaudi_dp2/llama3-8b/r8/point.yaml} (100%)
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/report.txt
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/run_metadata.json
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/sweep_distributions.csv
 create mode 100644 test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/sweep_summary.csv
 rename test_submissions/sub_f/{systems/sys_gaudi_dp2.json => sys_gaudi_dp2/system_desc.json} (100%)
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/accuracy/accuracy_result.json
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_1024 => sys_8gpu_vllm/llama3-70b/r1024}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_1024 => sys_8gpu_vllm/llama3-70b/r1024}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/points/point_1024.yaml => sys_8gpu_vllm/llama3-70b/r1024/point.yaml} (100%)
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/report.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/run_metadata.json
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/accuracy/accuracy_result.json
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_128 => sys_8gpu_vllm/llama3-70b/r128}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_128 => sys_8gpu_vllm/llama3-70b/r128}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/points/point_128.yaml => sys_8gpu_vllm/llama3-70b/r128/point.yaml} (100%)
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/report.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/run_metadata.json
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/accuracy/accuracy_result.json
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_1536 => sys_8gpu_vllm/llama3-70b/r1536}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_1536 => sys_8gpu_vllm/llama3-70b/r1536}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/points/point_1536.yaml => sys_8gpu_vllm/llama3-70b/r1536/point.yaml} (100%)
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/report.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/run_metadata.json
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/accuracy/accuracy_result.json
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_192 => sys_8gpu_vllm/llama3-70b/r192}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_192 => sys_8gpu_vllm/llama3-70b/r192}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/points/point_192.yaml => sys_8gpu_vllm/llama3-70b/r192/point.yaml} (100%)
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/report.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/run_metadata.json
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/accuracy/accuracy_result.json
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_2048 => sys_8gpu_vllm/llama3-70b/r2048}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_2048 => sys_8gpu_vllm/llama3-70b/r2048}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/points/point_2048.yaml => sys_8gpu_vllm/llama3-70b/r2048/point.yaml} (100%)
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/report.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/run_metadata.json
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/accuracy/accuracy_result.json
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_256 => sys_8gpu_vllm/llama3-70b/r256}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_256 => sys_8gpu_vllm/llama3-70b/r256}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/points/point_256.yaml => sys_8gpu_vllm/llama3-70b/r256/point.yaml} (100%)
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/report.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/run_metadata.json
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/accuracy/accuracy_result.json
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_384 => sys_8gpu_vllm/llama3-70b/r384}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_384 => sys_8gpu_vllm/llama3-70b/r384}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/points/point_384.yaml => sys_8gpu_vllm/llama3-70b/r384/point.yaml} (100%)
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/report.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/run_metadata.json
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/accuracy/accuracy_result.json
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_512 => sys_8gpu_vllm/llama3-70b/r512}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_512 => sys_8gpu_vllm/llama3-70b/r512}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/points/point_512.yaml => sys_8gpu_vllm/llama3-70b/r512/point.yaml} (100%)
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/report.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/run_metadata.json
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/accuracy/accuracy_result.json
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_64 => sys_8gpu_vllm/llama3-70b/r64}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_64 => sys_8gpu_vllm/llama3-70b/r64}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/points/point_64.yaml => sys_8gpu_vllm/llama3-70b/r64/point.yaml} (100%)
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/report.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/run_metadata.json
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/accuracy/accuracy_result.json
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_768 => sys_8gpu_vllm/llama3-70b/r768}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/results/point_768 => sys_8gpu_vllm/llama3-70b/r768}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_g/{pareto/sys_8gpu_vllm/llama3-70b/points/point_768.yaml => sys_8gpu_vllm/llama3-70b/r768/point.yaml} (100%)
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/report.txt
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/run_metadata.json
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/sweep_distributions.csv
 create mode 100644 test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/sweep_summary.csv
 rename test_submissions/sub_g/{systems/sys_8gpu_vllm.json => sys_8gpu_vllm/system_desc.json} (100%)
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/accuracy/accuracy_result.json
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_1024 => sys_8gpu_sglang/llama3-70b/r1024}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_1024 => sys_8gpu_sglang/llama3-70b/r1024}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/points/point_1024.yaml => sys_8gpu_sglang/llama3-70b/r1024/point.yaml} (100%)
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/report.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/run_metadata.json
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/accuracy/accuracy_result.json
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_128 => sys_8gpu_sglang/llama3-70b/r128}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_128 => sys_8gpu_sglang/llama3-70b/r128}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/points/point_128.yaml => sys_8gpu_sglang/llama3-70b/r128/point.yaml} (100%)
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/report.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/run_metadata.json
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/accuracy/accuracy_result.json
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_1536 => sys_8gpu_sglang/llama3-70b/r1536}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_1536 => sys_8gpu_sglang/llama3-70b/r1536}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/points/point_1536.yaml => sys_8gpu_sglang/llama3-70b/r1536/point.yaml} (100%)
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/report.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/run_metadata.json
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/accuracy/accuracy_result.json
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_192 => sys_8gpu_sglang/llama3-70b/r192}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_192 => sys_8gpu_sglang/llama3-70b/r192}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/points/point_192.yaml => sys_8gpu_sglang/llama3-70b/r192/point.yaml} (100%)
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/report.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/run_metadata.json
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/accuracy/accuracy_result.json
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_2048 => sys_8gpu_sglang/llama3-70b/r2048}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_2048 => sys_8gpu_sglang/llama3-70b/r2048}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/points/point_2048.yaml => sys_8gpu_sglang/llama3-70b/r2048/point.yaml} (100%)
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/report.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/run_metadata.json
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/accuracy/accuracy_result.json
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_256 => sys_8gpu_sglang/llama3-70b/r256}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_256 => sys_8gpu_sglang/llama3-70b/r256}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/points/point_256.yaml => sys_8gpu_sglang/llama3-70b/r256/point.yaml} (100%)
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/report.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/run_metadata.json
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/accuracy/accuracy_result.json
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_384 => sys_8gpu_sglang/llama3-70b/r384}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_384 => sys_8gpu_sglang/llama3-70b/r384}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/points/point_384.yaml => sys_8gpu_sglang/llama3-70b/r384/point.yaml} (100%)
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/report.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/run_metadata.json
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/accuracy/accuracy_result.json
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_512 => sys_8gpu_sglang/llama3-70b/r512}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_512 => sys_8gpu_sglang/llama3-70b/r512}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/points/point_512.yaml => sys_8gpu_sglang/llama3-70b/r512/point.yaml} (100%)
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/report.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/run_metadata.json
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/accuracy/accuracy_result.json
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_64 => sys_8gpu_sglang/llama3-70b/r64}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_64 => sys_8gpu_sglang/llama3-70b/r64}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/points/point_64.yaml => sys_8gpu_sglang/llama3-70b/r64/point.yaml} (100%)
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/report.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/run_metadata.json
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/accuracy/accuracy_result.json
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_768 => sys_8gpu_sglang/llama3-70b/r768}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/results/point_768 => sys_8gpu_sglang/llama3-70b/r768}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_h/{pareto/sys_8gpu_sglang/llama3-70b/points/point_768.yaml => sys_8gpu_sglang/llama3-70b/r768/point.yaml} (100%)
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/report.txt
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/run_metadata.json
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/sweep_distributions.csv
 create mode 100644 test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/sweep_summary.csv
 rename test_submissions/sub_h/{systems/sys_8gpu_sglang.json => sys_8gpu_sglang/system_desc.json} (100%)
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/accuracy/accuracy_result.json
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_1 => sys_h200_8gpu/deepseek-r1/r1}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_1 => sys_h200_8gpu/deepseek-r1/r1}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/points/point_1.yaml => sys_h200_8gpu/deepseek-r1/r1/point.yaml} (100%)
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/report.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/run_metadata.json
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/accuracy/accuracy_result.json
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_128 => sys_h200_8gpu/deepseek-r1/r128}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_128 => sys_h200_8gpu/deepseek-r1/r128}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/points/point_128.yaml => sys_h200_8gpu/deepseek-r1/r128/point.yaml} (100%)
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/report.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/run_metadata.json
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/accuracy/accuracy_result.json
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_16 => sys_h200_8gpu/deepseek-r1/r16}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_16 => sys_h200_8gpu/deepseek-r1/r16}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/points/point_16.yaml => sys_h200_8gpu/deepseek-r1/r16/point.yaml} (100%)
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/report.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/run_metadata.json
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/accuracy/accuracy_result.json
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_2 => sys_h200_8gpu/deepseek-r1/r2}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_2 => sys_h200_8gpu/deepseek-r1/r2}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/points/point_2.yaml => sys_h200_8gpu/deepseek-r1/r2/point.yaml} (100%)
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/report.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/run_metadata.json
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/accuracy/accuracy_result.json
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_256 => sys_h200_8gpu/deepseek-r1/r256}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_256 => sys_h200_8gpu/deepseek-r1/r256}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/points/point_256.yaml => sys_h200_8gpu/deepseek-r1/r256/point.yaml} (100%)
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/report.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/run_metadata.json
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/accuracy/accuracy_result.json
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_32 => sys_h200_8gpu/deepseek-r1/r32}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_32 => sys_h200_8gpu/deepseek-r1/r32}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/points/point_32.yaml => sys_h200_8gpu/deepseek-r1/r32/point.yaml} (100%)
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/report.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/run_metadata.json
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/accuracy/accuracy_result.json
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_4 => sys_h200_8gpu/deepseek-r1/r4}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_4 => sys_h200_8gpu/deepseek-r1/r4}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/points/point_4.yaml => sys_h200_8gpu/deepseek-r1/r4/point.yaml} (100%)
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/report.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/run_metadata.json
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/accuracy/accuracy_result.json
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_512 => sys_h200_8gpu/deepseek-r1/r512}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_512 => sys_h200_8gpu/deepseek-r1/r512}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/points/point_512.yaml => sys_h200_8gpu/deepseek-r1/r512/point.yaml} (100%)
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/report.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/run_metadata.json
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/accuracy/accuracy_result.json
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_64 => sys_h200_8gpu/deepseek-r1/r64}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_64 => sys_h200_8gpu/deepseek-r1/r64}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/points/point_64.yaml => sys_h200_8gpu/deepseek-r1/r64/point.yaml} (100%)
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/report.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/run_metadata.json
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/accuracy/accuracy_result.json
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_8 => sys_h200_8gpu/deepseek-r1/r8}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/results/point_8 => sys_h200_8gpu/deepseek-r1/r8}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_i/{pareto/sys_h200_8gpu/deepseek-r1/points/point_8.yaml => sys_h200_8gpu/deepseek-r1/r8/point.yaml} (100%)
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/report.txt
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/run_metadata.json
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/sweep_distributions.csv
 create mode 100644 test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/sweep_summary.csv
 rename test_submissions/sub_i/{systems/sys_h200_8gpu.json => sys_h200_8gpu/system_desc.json} (100%)
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/accuracy/accuracy_result.json
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_1024 => sys_gb300_72gpu/deepseek-r1/r1024}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_1024 => sys_gb300_72gpu/deepseek-r1/r1024}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/points/point_1024.yaml => sys_gb300_72gpu/deepseek-r1/r1024/point.yaml} (100%)
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/report.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/run_metadata.json
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/accuracy/accuracy_result.json
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_128 => sys_gb300_72gpu/deepseek-r1/r128}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_128 => sys_gb300_72gpu/deepseek-r1/r128}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/points/point_128.yaml => sys_gb300_72gpu/deepseek-r1/r128/point.yaml} (100%)
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/report.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/run_metadata.json
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/accuracy/accuracy_result.json
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_16384 => sys_gb300_72gpu/deepseek-r1/r16384}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_16384 => sys_gb300_72gpu/deepseek-r1/r16384}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/points/point_16384.yaml => sys_gb300_72gpu/deepseek-r1/r16384/point.yaml} (100%)
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/report.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/run_metadata.json
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/accuracy/accuracy_result.json
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_2048 => sys_gb300_72gpu/deepseek-r1/r2048}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_2048 => sys_gb300_72gpu/deepseek-r1/r2048}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/points/point_2048.yaml => sys_gb300_72gpu/deepseek-r1/r2048/point.yaml} (100%)
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/report.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/run_metadata.json
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/accuracy/accuracy_result.json
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_256 => sys_gb300_72gpu/deepseek-r1/r256}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_256 => sys_gb300_72gpu/deepseek-r1/r256}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/points/point_256.yaml => sys_gb300_72gpu/deepseek-r1/r256/point.yaml} (100%)
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/report.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/run_metadata.json
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/accuracy/accuracy_result.json
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_32 => sys_gb300_72gpu/deepseek-r1/r32}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_32 => sys_gb300_72gpu/deepseek-r1/r32}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/points/point_32.yaml => sys_gb300_72gpu/deepseek-r1/r32/point.yaml} (100%)
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/report.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/run_metadata.json
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/accuracy/accuracy_result.json
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_4096 => sys_gb300_72gpu/deepseek-r1/r4096}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_4096 => sys_gb300_72gpu/deepseek-r1/r4096}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/points/point_4096.yaml => sys_gb300_72gpu/deepseek-r1/r4096/point.yaml} (100%)
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/report.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/run_metadata.json
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/accuracy/accuracy_result.json
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_512 => sys_gb300_72gpu/deepseek-r1/r512}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_512 => sys_gb300_72gpu/deepseek-r1/r512}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/points/point_512.yaml => sys_gb300_72gpu/deepseek-r1/r512/point.yaml} (100%)
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/report.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/run_metadata.json
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/accuracy/accuracy_result.json
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_64 => sys_gb300_72gpu/deepseek-r1/r64}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_64 => sys_gb300_72gpu/deepseek-r1/r64}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/points/point_64.yaml => sys_gb300_72gpu/deepseek-r1/r64/point.yaml} (100%)
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/report.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/run_metadata.json
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/accuracy/accuracy.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/accuracy/accuracy_result.json
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_8192 => sys_gb300_72gpu/deepseek-r1/r8192}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/results/point_8192 => sys_gb300_72gpu/deepseek-r1/r8192}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/sub_j/{pareto/sys_gb300_72gpu/deepseek-r1/points/point_8192.yaml => sys_gb300_72gpu/deepseek-r1/r8192/point.yaml} (100%)
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/report.txt
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/run_metadata.json
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/src/vllm/.gitkeep
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/sweep_distributions.csv
 create mode 100644 test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/sweep_summary.csv
 rename test_submissions/sub_j/{systems/sys_gb300_72gpu.json => sys_gb300_72gpu/system_desc.json} (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b => acme_h100x8_001/llama3-70b/r1000}/accuracy/accuracy.txt (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b => acme_h100x8_001/llama3-70b/r1000}/accuracy/accuracy_result.json (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/results/point_1000 => acme_h100x8_001/llama3-70b/r1000}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/results/point_1000 => acme_h100x8_001/llama3-70b/r1000}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/points/point_1000.yaml => acme_h100x8_001/llama3-70b/r1000/point.yaml} (100%)
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/report.txt
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/run_metadata.json
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/src/vllm/.gitkeep
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/accuracy/accuracy.txt
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/accuracy/accuracy_result.json
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/results/point_16 => acme_h100x8_001/llama3-70b/r16}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/results/point_16 => acme_h100x8_001/llama3-70b/r16}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/points/point_16.yaml => acme_h100x8_001/llama3-70b/r16/point.yaml} (100%)
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/report.txt
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/run_metadata.json
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/src/vllm/.gitkeep
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/accuracy/accuracy.txt
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/accuracy/accuracy_result.json
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/results/point_256 => acme_h100x8_001/llama3-70b/r256}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/results/point_256 => acme_h100x8_001/llama3-70b/r256}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/points/point_256.yaml => acme_h100x8_001/llama3-70b/r256/point.yaml} (100%)
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/report.txt
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/run_metadata.json
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/src/vllm/.gitkeep
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/accuracy/accuracy.txt
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/accuracy/accuracy_result.json
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/results/point_38 => acme_h100x8_001/llama3-70b/r38}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/results/point_38 => acme_h100x8_001/llama3-70b/r38}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/points/point_38.yaml => acme_h100x8_001/llama3-70b/r38/point.yaml} (100%)
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/report.txt
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/run_metadata.json
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/src/vllm/.gitkeep
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/accuracy/accuracy.txt
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/accuracy/accuracy_result.json
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/results/point_512 => acme_h100x8_001/llama3-70b/r512}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/results/point_512 => acme_h100x8_001/llama3-70b/r512}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/points/point_512.yaml => acme_h100x8_001/llama3-70b/r512/point.yaml} (100%)
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/report.txt
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/run_metadata.json
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/src/vllm/.gitkeep
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/accuracy/accuracy.txt
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/accuracy/accuracy_result.json
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/results/point_768 => acme_h100x8_001/llama3-70b/r768}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/results/point_768 => acme_h100x8_001/llama3-70b/r768}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/points/point_768.yaml => acme_h100x8_001/llama3-70b/r768/point.yaml} (100%)
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/report.txt
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/run_metadata.json
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/src/vllm/.gitkeep
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/accuracy/accuracy.txt
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/accuracy/accuracy_result.json
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/results/point_88 => acme_h100x8_001/llama3-70b/r88}/mlperf_endpoints_log_detail.json (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/results/point_88 => acme_h100x8_001/llama3-70b/r88}/mlperf_endpoints_log_summary.json (100%)
 rename test_submissions/valid_standardized/{pareto/acme_h100x8_001/llama3-70b/points/point_88.yaml => acme_h100x8_001/llama3-70b/r88/point.yaml} (100%)
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/report.txt
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/run_metadata.json
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/src/vllm/.gitkeep
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/sweep_distributions.csv
 create mode 100644 test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/sweep_summary.csv
 rename test_submissions/valid_standardized/{systems/acme_h100x8_001.json => acme_h100x8_001/system_desc.json} (100%)

diff --git a/src/endpoints_submission_cli/submissions/builder.py b/src/endpoints_submission_cli/submissions/builder.py
index aa9f944..93d1f83 100644
--- a/src/endpoints_submission_cli/submissions/builder.py
+++ b/src/endpoints_submission_cli/submissions/builder.py
@@ -6,14 +6,23 @@
 into the SubmissionChecker-compatible layout:
 
     <org>/
-      systems/<system_id>.json
-      pareto/<system_id>/<model>/points/point_<concurrency>.yaml
-      pareto/<system_id>/<model>/results/point_<concurrency>/
-          mlperf_endpoints_log_summary.json
-          mlperf_endpoints_log_detail.json
-      pareto/<system_id>/<model>/accuracy/
-          accuracy_result.json
-          accuracy.txt
+      <system_id>/
+        system_desc.json
+        <model>/
+          sweep_summary.csv
+          sweep_distributions.csv
+          r<concurrency>/
+            point.yaml
+            mlperf_endpoints_log_summary.json
+            mlperf_endpoints_log_detail.json
+            run_metadata.json
+            report.txt
+            accuracy/
+              accuracy_result.json
+              accuracy.txt
+            src/
+              <implementation>/
+                (endpoint interface code)
 """
 
 from __future__ import annotations
@@ -88,16 +97,14 @@ def build_submission_folder(
     for (system_id, model), runs in groups.items():
         all_system_runs = runs_by_system[system_id]
         if system_id not in written_systems:
+            max_concurrency = max(_extract_concurrency(r["config"]) for r in all_system_runs)
             _write_system_description(
-                submission_dir, system_id, model, all_system_runs, division
+                submission_dir, system_id, model, all_system_runs, division, max_concurrency
             )
             written_systems.add(system_id)
         max_concurrency = max(_extract_concurrency(r["config"]) for r in all_system_runs)
-        _write_pareto_entries(submission_dir, system_id, model, runs, max_concurrency)
-        _write_accuracy_placeholders(submission_dir, system_id, model)
-
-    if _normalize_division(division) == "Standardized":
-        (submission_dir / "src").mkdir(exist_ok=True)
+        _write_run_entries(submission_dir, system_id, model, runs, max_concurrency)
+        _write_model_sweep_stubs(submission_dir, system_id, model)
 
     return submission_dir
 
@@ -224,13 +231,10 @@ def _write_system_description(
     model: str,
     runs: list[dict[str, Any]],
     division: str,
+    max_concurrency: int,
 ) -> None:
-    systems_dir = submission_dir / "systems"
-    systems_dir.mkdir(parents=True, exist_ok=True)
-
-    # Derive max_supported_concurrency from the highest concurrency across all runs
-    concurrencies = [_extract_concurrency(r["config"]) for r in runs]
-    max_concurrency = max(concurrencies) if concurrencies else 64
+    system_dir = submission_dir / system_id
+    system_dir.mkdir(parents=True, exist_ok=True)
 
     si = runs[0]["system_info"]
     cfg = runs[0]["config"]
@@ -281,12 +285,12 @@ def _write_system_description(
     ):
         system_desc["host_processor_core_count"] = 1
 
-    (systems_dir / f"{system_id}.json").write_text(
+    (system_dir / "system_desc.json").write_text(
         json.dumps(system_desc, indent=2), encoding="utf-8"
     )
 
 
-def _write_pareto_entries(
+def _write_run_entries(
     submission_dir: Path,
     system_id: str,
     model: str,
@@ -299,13 +303,10 @@ def _write_pareto_entries(
 
     for run in runs:
         concurrency = _extract_concurrency(run["config"])
-        model_dir = submission_dir / "pareto" / system_id / model
-        points_dir = model_dir / "points"
-        result_dir = model_dir / "results" / f"point_{concurrency}"
-        points_dir.mkdir(parents=True, exist_ok=True)
-        result_dir.mkdir(parents=True, exist_ok=True)
+        run_dir = submission_dir / system_id / model / f"r{concurrency}"
+        run_dir.mkdir(parents=True, exist_ok=True)
 
-        # Build point YAML from config.yaml + runtime_settings.json
+        # Build point.yaml from config.yaml + runtime_settings.json
         cfg_settings = run["config"].get("settings", {}) or {}
         load_pattern = cfg_settings.get("load_pattern", {}) or {}
         rt_json = run.get("runtime_settings", {}) or {}
@@ -326,37 +327,65 @@ def _write_pareto_entries(
             "dataset": dataset_name,
             "runtime_settings": runtime_settings_out,
         }
-        (points_dir / f"point_{concurrency}.yaml").write_text(
+        (run_dir / "point.yaml").write_text(
             yaml.dump(point_cfg, default_flow_style=False), encoding="utf-8"
         )
 
-        (result_dir / "mlperf_endpoints_log_summary.json").write_text(
+        (run_dir / "mlperf_endpoints_log_summary.json").write_text(
             json.dumps(run["result_summary"], indent=2), encoding="utf-8"
         )
-        (result_dir / "mlperf_endpoints_log_detail.json").write_text(
+        (run_dir / "mlperf_endpoints_log_detail.json").write_text(
             "{}", encoding="utf-8"
         )
-        (result_dir / "system_desc.json").write_text(
-            json.dumps(run["system_info"], indent=2), encoding="utf-8"
+        (run_dir / "run_metadata.json").write_text(
+            json.dumps(
+                {
+                    "serving_framework": run["system_info"].get("framework", ""),
+                    "parallelism": {},
+                    "precision": "bfloat16",
+                },
+                indent=2,
+            ),
+            encoding="utf-8",
+        )
+        (run_dir / "report.txt").write_text(
+            f"Run r{concurrency} — auto-generated placeholder\n", encoding="utf-8"
+        )
+
+        # src/<impl>/ stub — submitter fills in endpoint interface code
+        src_dir = run_dir / "src" / "vllm"
+        src_dir.mkdir(parents=True, exist_ok=True)
+        (src_dir / ".gitkeep").write_text("", encoding="utf-8")
+
+        # Per-run accuracy placeholder
+        accuracy_dir = run_dir / "accuracy"
+        accuracy_dir.mkdir(exist_ok=True)
+        (accuracy_dir / "accuracy.txt").write_text("Accuracy pending\n", encoding="utf-8")
+        placeholder = {
+            "metric": "rouge1",
+            "score": 0.0,
+            "quality_target": 0.0,
+            "passed": True,
+        }
+        (accuracy_dir / "accuracy_result.json").write_text(
+            json.dumps(placeholder, indent=2), encoding="utf-8"
         )
 
 
-def _write_accuracy_placeholders(
+def _write_model_sweep_stubs(
     submission_dir: Path,
     system_id: str,
     model: str,
 ) -> None:
-    accuracy_dir = submission_dir / "pareto" / system_id / model / "accuracy"
-    accuracy_dir.mkdir(parents=True, exist_ok=True)
-    (accuracy_dir / "accuracy.txt").write_text("Accuracy pending\n", encoding="utf-8")
-    placeholder = {
-        "metric": "rouge1",
-        "score": 0.0,
-        "quality_target": 0.0,
-        "passed": True,
-    }
-    (accuracy_dir / "accuracy_result.json").write_text(
-        json.dumps(placeholder, indent=2), encoding="utf-8"
+    model_dir = submission_dir / system_id / model
+    model_dir.mkdir(parents=True, exist_ok=True)
+    (model_dir / "sweep_summary.csv").write_text(
+        "concurrency,qps,ttft_p50_ms,ttft_p95_ms,tpot_p50_ms,system_tps\n",
+        encoding="utf-8",
+    )
+    (model_dir / "sweep_distributions.csv").write_text(
+        "concurrency,percentile,ttft_ms,tpot_ms,output_tokens\n",
+        encoding="utf-8",
     )
 
 
diff --git a/src/endpoints_submission_cli/submissions/github.py b/src/endpoints_submission_cli/submissions/github.py
index 6804bf8..fc9d57b 100644
--- a/src/endpoints_submission_cli/submissions/github.py
+++ b/src/endpoints_submission_cli/submissions/github.py
@@ -10,6 +10,7 @@
 from __future__ import annotations
 
 import os
+import re
 import shutil
 import subprocess
 from pathlib import Path
@@ -194,55 +195,62 @@ def prepare_pr_branch_merge(
     repo_org_dir = repo_dir / submission_dir.name  # e.g. repo_dir / "NVIDIA"
 
     if repo_org_dir.exists():
-        fresh_pareto = submission_dir / "pareto"
-        repo_pareto = repo_org_dir / "pareto"
-
-        for fresh_model_dir in fresh_pareto.glob("*/*"):  # <system_id>/<model>
-            rel = fresh_model_dir.relative_to(fresh_pareto)
-            repo_model_dir = repo_pareto / rel
-            repo_model_dir.mkdir(parents=True, exist_ok=True)
-
-            # points/ and accuracy/ — replace entirely from fresh build
-            for subdir_name in ("points", "accuracy"):
-                dest = repo_model_dir / subdir_name
-                if dest.exists():
-                    shutil.rmtree(dest)
-                src = fresh_model_dir / subdir_name
-                if src.exists():
-                    shutil.copytree(src, dest)
-
-            # results/ — surgical per-point update
-            fresh_results = fresh_model_dir / "results"
-            repo_results = repo_model_dir / "results"
-            if fresh_results.exists():
-                repo_results.mkdir(exist_ok=True)
-                # Remove point dirs no longer present in fresh build
-                fresh_point_names = {p.name for p in fresh_results.iterdir() if p.is_dir()}
-                for repo_point in list(repo_results.iterdir()):
-                    if repo_point.is_dir() and repo_point.name not in fresh_point_names:
-                        shutil.rmtree(repo_point)
-                # Update each point: replace log files, preserve system_desc.json
-                for fresh_point in fresh_results.iterdir():
-                    if not fresh_point.is_dir():
+        # Iterate system directories in fresh build (identified by system_desc.json)
+        for fresh_sys_dir in sorted(submission_dir.iterdir()):
+            if not fresh_sys_dir.is_dir():
+                continue
+            repo_sys_dir = repo_org_dir / fresh_sys_dir.name
+            repo_sys_dir.mkdir(parents=True, exist_ok=True)
+
+            # system_desc.json: preserve PR version; seed from fresh build if absent
+            repo_sysdesc = repo_sys_dir / "system_desc.json"
+            fresh_sysdesc = fresh_sys_dir / "system_desc.json"
+            if not repo_sysdesc.exists() and fresh_sysdesc.exists():
+                shutil.copy2(fresh_sysdesc, repo_sysdesc)
+
+            for fresh_model_dir in sorted(fresh_sys_dir.iterdir()):
+                if not fresh_model_dir.is_dir():
+                    continue
+                repo_model_dir = repo_sys_dir / fresh_model_dir.name
+                repo_model_dir.mkdir(parents=True, exist_ok=True)
+
+                # sweep CSVs — always replace from fresh build
+                for csv_name in ("sweep_summary.csv", "sweep_distributions.csv"):
+                    src_csv = fresh_model_dir / csv_name
+                    if src_csv.exists():
+                        shutil.copy2(src_csv, repo_model_dir / csv_name)
+
+                # r<N>/ run dirs — surgical per-run update
+                fresh_run_names = {
+                    d.name
+                    for d in fresh_model_dir.iterdir()
+                    if d.is_dir() and re.match(r"^r\d+$", d.name)
+                }
+                # Remove run dirs no longer present in fresh build
+                if repo_model_dir.exists():
+                    for repo_run in list(repo_model_dir.iterdir()):
+                        if repo_run.is_dir() and re.match(r"^r\d+$", repo_run.name) and repo_run.name not in fresh_run_names:
+                            shutil.rmtree(repo_run)
+                # Update each run dir
+                for fresh_run in sorted(fresh_model_dir.iterdir()):
+                    if not fresh_run.is_dir() or not re.match(r"^r\d+$", fresh_run.name):
                         continue
-                    repo_point = repo_results / fresh_point.name
-                    is_new_point = not repo_point.exists()
-                    repo_point.mkdir(exist_ok=True)
-                    for src_file in fresh_point.iterdir():
-                        if src_file.name != "system_desc.json":
-                            shutil.copy2(src_file, repo_point / src_file.name)
-                    # system_desc.json: preserve PR version; seed only for new points
-                    repo_sysdesc = repo_point / "system_desc.json"
-                    if is_new_point or not repo_sysdesc.exists():
-                        fresh_sysdesc = fresh_point / "system_desc.json"
-                        if fresh_sysdesc.exists():
-                            shutil.copy2(fresh_sysdesc, repo_sysdesc)
-            elif repo_results.exists():
-                shutil.rmtree(repo_results)
-
-        # systems/ — preserve PR version; seed from fresh build if absent
-        if not (repo_org_dir / "systems").exists():
-            shutil.copytree(submission_dir / "systems", repo_org_dir / "systems")
+                    repo_run = repo_model_dir / fresh_run.name
+                    is_new = not repo_run.exists()
+                    repo_run.mkdir(exist_ok=True)
+                    # Replace all items EXCEPT src/ (may have manually-authored endpoint config)
+                    for item in fresh_run.iterdir():
+                        dest = repo_run / item.name
+                        if item.name == "src":
+                            # Preserve from PR; seed only for new runs
+                            if is_new and not dest.exists():
+                                shutil.copytree(item, dest)
+                        elif item.is_dir():
+                            if dest.exists():
+                                shutil.rmtree(dest)
+                            shutil.copytree(item, dest)
+                        else:
+                            shutil.copy2(item, dest)
     else:
         # Org dir not yet on the PR branch — full copy (first push edge case).
         shutil.copytree(submission_dir, repo_org_dir)
diff --git a/src/submission_checker/checker.py b/src/submission_checker/checker.py
index 1ce7470..fefbd71 100644
--- a/src/submission_checker/checker.py
+++ b/src/submission_checker/checker.py
@@ -6,6 +6,7 @@
 
 from __future__ import annotations
 
+import re
 from typing import TYPE_CHECKING
 
 __all__ = ["SubmissionChecker"]
@@ -24,12 +25,13 @@
     PointSummary,
     Regions,
     Report,
+    RunDir,
     RuntimeSettings,
     Severity,
     SystemDescription,
     compute_regions,
 )
-from .models import ModelDir, SrcDir, SubmissionDir, SystemPareto
+from .models import ModelDir
 from .models import err as _err
 from .models import ok as _ok
 from .models import warn as _warn
@@ -41,9 +43,9 @@
 class SubmissionChecker:
     """Validates an MLPerf Endpoints submission directory against §9.1 rules.
 
-    The *submission_path* should be the submitting organisation's root directory,
-    which must contain ``systems/`` and ``pareto/`` subdirectories as specified
-    in §8.1.
+    The *submission_path* should be the submitting organisation's root directory.
+    System directories are discovered by the presence of ``system_desc.json``
+    inside each immediate subdirectory (§8.1).
 
     Args:
         submission_path: Root directory of the submission to validate.
@@ -87,28 +89,24 @@ def run(self) -> Report:
             _ok("path-exists", "Submission path exists", self.submission_path, "#1")
         )
 
-        submission_dir = SubmissionDir(root=self.submission_path)
-        report.results.extend(submission_dir._check_results)
-        if any(r.severity == Severity.ERROR for r in submission_dir._check_results):
-            return report
-
-        systems_dir = submission_dir.systems_dir
-        pareto_dir = submission_dir.pareto_dir
-
-        system_jsons = sorted(systems_dir.glob("*.json"))
-        if not system_jsons:
+        system_dirs = [
+            d
+            for d in sorted(self.submission_path.iterdir())
+            if d.is_dir() and (d / "system_desc.json").exists()
+        ]
+        if not system_dirs:
             report.results.append(
                 _err(
-                    "system-description-present",
-                    "No *.json files found in systems/",
-                    systems_dir,
+                    "system-dir-present",
+                    "No system directories found (each system directory must contain system_desc.json)",
+                    self.submission_path,
                     "#1",
                 )
             )
             return report
 
-        for system_json in system_jsons:
-            report.results.extend(self._check_system(system_json, pareto_dir))
+        for system_dir in system_dirs:
+            report.results.extend(self._check_system(system_dir))
 
         return report
 
@@ -116,9 +114,10 @@ def run(self) -> Report:
     # Per-system orchestration
     # ------------------------------------------------------------------
 
-    def _check_system(self, system_json: Path, pareto_dir: Path) -> list[CheckResult]:
+    def _check_system(self, system_dir: Path) -> list[CheckResult]:
         results: list[CheckResult] = []
-        system_id = system_json.stem
+        system_name = system_dir.name
+        system_json = system_dir / "system_desc.json"
 
         system_desc, load_results = load_system_description(system_json)
         results.extend(load_results)
@@ -127,7 +126,7 @@ def _check_system(self, system_json: Path, pareto_dir: Path) -> list[CheckResult
         results.append(
             _ok(
                 "system-description-valid",
-                f"System description valid: {system_id}",
+                f"System description valid: {system_name}",
                 system_json,
                 "#1",
             )
@@ -144,28 +143,22 @@ def _check_system(self, system_json: Path, pareto_dir: Path) -> list[CheckResult
             results.append(_err("region-computation", str(exc), system_json, "#7"))
             return results
 
-        src = SrcDir(root=self.submission_path, division=system_desc.division)
-        results.extend(src._check_results)
-
-        system_pareto = SystemPareto(pareto_dir=pareto_dir, system_id=system_id)
-        results.extend(system_pareto._check_results)
-        if any(r.severity == Severity.ERROR for r in system_pareto._check_results):
-            return results
-        system_pareto_dir = system_pareto.system_dir
-        model_dirs = [d for d in sorted(system_pareto_dir.iterdir()) if d.is_dir()]
+        model_dirs = [
+            d for d in sorted(system_dir.iterdir()) if d.is_dir() and d.name != "docs"
+        ]
         if not model_dirs:
             results.append(
                 _err(
                     "benchmark-model-dir",
-                    f"No benchmark-model directories in pareto/{system_id}/",
-                    system_pareto_dir,
+                    f"No benchmark-model directories in {system_name}/",
+                    system_dir,
                     "#1",
                 )
             )
             return results
 
         for model_dir in model_dirs:
-            results.extend(self._check_model(system_id, system_desc, regions, model_dir))
+            results.extend(self._check_model(system_name, system_desc, regions, model_dir))
 
         return results
 
@@ -175,32 +168,26 @@ def _check_system(self, system_json: Path, pareto_dir: Path) -> list[CheckResult
 
     def _check_model(
         self,
-        system_id: str,
+        system_name: str,
         system_desc: SystemDescription,
         regions: Regions,
         model_dir: Path,
     ) -> list[CheckResult]:
         results: list[CheckResult] = []
-        benchmark_model = model_dir.name
+        model_name = model_dir.name
 
         model_structure = ModelDir(
-            root=model_dir, system_id=system_id, benchmark_model=benchmark_model
+            root=model_dir, system_name=system_name, model_name=model_name
         )
         results.extend(model_structure._check_results)
-        if any(r.severity == Severity.ERROR for r in model_structure._check_results):
-            return results
 
-        points_dir = model_structure.points_dir
-        results_dir = model_structure.results_dir
-        accuracy_dir = model_structure.accuracy_dir
-
-        point_yamls = sorted(points_dir.glob("point_*.yaml"))
-        if not point_yamls:
+        run_dirs = model_structure.run_dirs
+        if not run_dirs:
             results.append(
                 _err(
                     "measurement-points-present",
-                    f"No point_*.yaml files in {points_dir.relative_to(self.submission_path)}",
-                    points_dir,
+                    f"No r<N>/ run directories in {system_name}/{model_name}/",
+                    model_dir,
                     "#1",
                 )
             )
@@ -208,42 +195,51 @@ def _check_model(
 
         valid_points: list[tuple[Path, PointConfig]] = []
         loaded_points: list[tuple[PointConfig, PointSummary]] = []
+        accuracy_results = []
 
-        for yaml_path in point_yamls:
+        for run_dir in run_dirs:
+            run_structure = RunDir(
+                root=run_dir, system_name=system_name, model_name=model_name
+            )
+            results.extend(run_structure._check_results)
+            if any(r.severity == Severity.ERROR for r in run_structure._check_results):
+                # Still try to load what we can for coverage checks
+                pass
+
+            point_yaml = run_dir / "point.yaml"
             config, config_results = load_point_config(
-                yaml_path, context={"regions": regions, "yaml_path": yaml_path}
+                point_yaml, context={"regions": regions, "yaml_path": point_yaml}
             )
             results.extend(config_results)
             if config is None:
                 continue
 
-            # filename-concurrency consistency warning
+            # dir-name concurrency consistency warning
             try:
-                fname_concurrency = int(yaml_path.stem.split("_")[1])
-                if fname_concurrency != config.concurrency:
+                dir_concurrency = int(run_dir.name[1:])  # strip leading 'r'
+                if dir_concurrency != config.concurrency:
                     results.append(
                         _warn(
                             "point-filename-concurrency",
-                            f"{yaml_path.name}: filename concurrency {fname_concurrency}"
+                            f"{run_dir.name}/: directory concurrency {dir_concurrency}"
                             f" ≠ declared {config.concurrency}",
-                            yaml_path,
+                            point_yaml,
                             "#1",
                         )
                     )
-            except (IndexError, ValueError):
+            except ValueError:
                 pass
 
-            valid_points.append((yaml_path, config))
+            valid_points.append((point_yaml, config))
 
-            point_result_dir = results_dir / f"point_{config.concurrency}"
-            summary_path = point_result_dir / "mlperf_endpoints_log_summary.json"
-            detail_path = point_result_dir / "mlperf_endpoints_log_detail.json"
+            summary_path = run_dir / "mlperf_endpoints_log_summary.json"
+            detail_path = run_dir / "mlperf_endpoints_log_detail.json"
 
             if not summary_path.exists():
                 results.append(
                     _err(
                         "result-file-present",
-                        f"Missing result log for point_{config.concurrency}:"
+                        f"Missing result log for {run_dir.name}:"
                         f" {summary_path.relative_to(self.submission_path)}",
                         summary_path,
                         "#1",
@@ -255,7 +251,7 @@ def _check_model(
                 results.append(
                     _err(
                         "result-detail-present",
-                        f"Missing detail log for point_{config.concurrency}:"
+                        f"Missing detail log for {run_dir.name}:"
                         f" {detail_path.relative_to(self.submission_path)}",
                         detail_path,
                         "#1",
@@ -269,40 +265,42 @@ def _check_model(
 
             # PointResult validates point-duration and metric-consistency
             point_result = PointResult.model_validate(
-                {"config": config, "summary": summary, "yaml_path": yaml_path},
+                {"config": config, "summary": summary, "yaml_path": point_yaml},
                 context={"regions": regions, "summary_path": summary_path},
             )
             results.extend(point_result._check_results)
             loaded_points.append((config, summary))
 
-        # Load accuracy
-        accuracy_result = None
-        txt_path = accuracy_dir / "accuracy.txt"
-        if not txt_path.exists():
-            results.append(_err("accuracy-file", "Missing accuracy/accuracy.txt", txt_path, "#15"))
-        json_path = accuracy_dir / "accuracy_result.json"
-        if not json_path.exists():
-            results.append(
-                _err("accuracy-file", "Missing accuracy/accuracy_result.json", json_path, "#15")
-            )
-        else:
-            accuracy_result, acc_results = load_accuracy_result(json_path)
-            results.extend(acc_results)
-            if any(r.severity == Severity.ERROR for r in acc_results):
-                accuracy_result = None
+            # Per-run accuracy
+            accuracy_dir = run_dir / "accuracy"
+            txt_path = accuracy_dir / "accuracy.txt"
+            if not txt_path.exists():
+                results.append(
+                    _err("accuracy-file", f"Missing accuracy/accuracy.txt in {run_dir.name}/", txt_path, "#15")
+                )
+            json_path = accuracy_dir / "accuracy_result.json"
+            if not json_path.exists():
+                results.append(
+                    _err("accuracy-file", f"Missing accuracy/accuracy_result.json in {run_dir.name}/", json_path, "#15")
+                )
+                accuracy_results.append(None)
+            else:
+                acc_result, acc_results = load_accuracy_result(json_path)
+                results.extend(acc_results)
+                if any(r.severity == Severity.ERROR for r in acc_results):
+                    acc_result = None
+                accuracy_results.append(acc_result)
 
         # ModelContext validates point-count, regional-coverage, config-consistency, accuracy-gate
         model_ctx = ModelContext(
-            system_id=system_id,
+            system_id=system_name,
             system_desc=system_desc,
             model_dir=model_dir,
             regions=regions,
-            points_dir=points_dir,
-            accuracy_dir=accuracy_dir,
-            all_point_count=len(point_yamls),
+            all_point_count=len(run_dirs),
             valid_points=valid_points,
             loaded_points=loaded_points,
-            accuracy_result=accuracy_result,
+            accuracy_results=accuracy_results,
         )
         results.extend(model_ctx._check_results)
 
diff --git a/src/submission_checker/models/__init__.py b/src/submission_checker/models/__init__.py
index 1d9932c..40d2b4a 100644
--- a/src/submission_checker/models/__init__.py
+++ b/src/submission_checker/models/__init__.py
@@ -13,7 +13,7 @@
 )
 from .regions import MIN_DURATION_MS, RegionBounds, Regions, classify_concurrency, compute_regions
 from .results import CheckResult, Report, Severity, err, ok, warn
-from .structure import ModelDir, SrcDir, SubmissionDir, SystemPareto
+from .structure import ModelDir, RunDir
 
 __all__ = [
     "AccuracyResult",
@@ -31,12 +31,10 @@
     "RegionBounds",
     "Regions",
     "Report",
+    "RunDir",
     "RuntimeSettings",
     "Severity",
-    "SrcDir",
-    "SubmissionDir",
     "SystemDescription",
-    "SystemPareto",
     "classify_concurrency",
     "compute_regions",
     "err",
diff --git a/src/submission_checker/models/aggregate/context.py b/src/submission_checker/models/aggregate/context.py
index de0393c..80cefaf 100644
--- a/src/submission_checker/models/aggregate/context.py
+++ b/src/submission_checker/models/aggregate/context.py
@@ -29,12 +29,10 @@ class ModelContext(BaseModel):
     system_desc: SystemDescription
     model_dir: Path
     regions: Regions
-    points_dir: Path
-    accuracy_dir: Path
     all_point_count: int
     valid_points: list[tuple[Path, PointConfig]]
     loaded_points: list[tuple[PointConfig, PointSummary]]
-    accuracy_result: AccuracyResult | None = None
+    accuracy_results: list[AccuracyResult | None] = []
 
     @model_validator(mode="after")
     def _check_point_count(self) -> ModelContext:
@@ -45,17 +43,17 @@ def _check_point_count(self) -> ModelContext:
                 err(
                     "point-count",
                     f"Only {n} measurement point(s) — minimum {_MIN_POINTS} required",
-                    self.points_dir,
+                    self.model_dir,
                     "#2, #8",
                 )
             )
         else:
             self._check_results.append(
-                ok("point-count", f"Point count OK: {n}", self.points_dir, "#2, #8")
+                ok("point-count", f"Point count OK: {n}", self.model_dir, "#2, #8")
             )
         if n > _MAX_POINTS:
             self._check_results.append(
-                err("point-cap", f"{n} points exceed the {_MAX_POINTS}-point cap", self.points_dir, "#2, #8")
+                err("point-cap", f"{n} points exceed the {_MAX_POINTS}-point cap", self.model_dir, "#2, #8")
             )
         return self
 
@@ -77,7 +75,7 @@ def _check_regional_coverage(self) -> ModelContext:
                     ok(
                         rule,
                         f"{label} region covered: {sorted(matching)} (range {bounds})",
-                        self.points_dir,
+                        self.model_dir,
                         "#3–6",
                     )
                 )
@@ -86,7 +84,7 @@ def _check_regional_coverage(self) -> ModelContext:
                     err(
                         rule,
                         f"No point in {label} region (concurrency {bounds})",
-                        self.points_dir,
+                        self.model_dir,
                         "#3–6",
                     )
                 )
@@ -140,28 +138,30 @@ def _check_config_consistency(self) -> ModelContext:
 
     @model_validator(mode="after")
     def _check_accuracy(self) -> ModelContext:
-        """§15: accuracy score must meet or exceed the benchmark quality_target."""
-        if self.accuracy_result is None:
-            return self  # file missing/invalid already reported by checker.py
-        accuracy = self.accuracy_result
-        json_path = self.accuracy_dir / "accuracy_result.json"
-        if accuracy.passed:
+        """§15: all per-run accuracy scores must meet or exceed their benchmark quality_target."""
+        valid_results = [r for r in self.accuracy_results if r is not None]
+        if not valid_results:
+            return self  # all missing/invalid already reported by checker.py
+        all_passed = all(r.passed for r in valid_results)
+        failed = [r for r in valid_results if not r.passed]
+        if all_passed:
             self._check_results.append(
                 ok(
                     "accuracy-gate",
-                    f"Accuracy gate PASSED: {accuracy.metric} = {accuracy.score:.4f}"
-                    f" ≥ target {accuracy.quality_target:.4f}",
-                    json_path,
+                    f"Accuracy gate PASSED for all {len(valid_results)} run(s)",
+                    self.model_dir,
                     "#15",
                 )
             )
         else:
+            detail = "; ".join(
+                f"{r.metric}={r.score:.4f} < {r.quality_target:.4f}" for r in failed
+            )
             self._check_results.append(
                 err(
                     "accuracy-gate",
-                    f"Accuracy gate FAILED: {accuracy.metric} = {accuracy.score:.4f}"
-                    f" < target {accuracy.quality_target:.4f}",
-                    json_path,
+                    f"Accuracy gate FAILED for {len(failed)} run(s): {detail}",
+                    self.model_dir,
                     "#15",
                 )
             )
diff --git a/src/submission_checker/models/structure.py b/src/submission_checker/models/structure.py
index 2e02829..0665f80 100644
--- a/src/submission_checker/models/structure.py
+++ b/src/submission_checker/models/structure.py
@@ -2,146 +2,124 @@
 
 from __future__ import annotations
 
+import re
 from pathlib import Path
 
 from pydantic import BaseModel, PrivateAttr, computed_field, model_validator
 
-from .file import Division
 from .results import CheckResult, err, ok
 
 
-class SubmissionDir(BaseModel):
-    """Validates the top-level submission directory: systems/ and pareto/ must exist."""
+class ModelDir(BaseModel):
+    """Validates a <model>/ directory: sweep CSVs and at least one r<N>/ run dir."""
 
     _check_results: list[CheckResult] = PrivateAttr(default_factory=list)
 
     root: Path
+    system_name: str
+    model_name: str
 
     @computed_field  # type: ignore[prop-decorator]
     @property
-    def systems_dir(self) -> Path:
-        return self.root / "systems"
-
-    @computed_field  # type: ignore[prop-decorator]
-    @property
-    def pareto_dir(self) -> Path:
-        return self.root / "pareto"
+    def run_dirs(self) -> list[Path]:
+        if not self.root.is_dir():
+            return []
+        return sorted(
+            d for d in self.root.iterdir() if d.is_dir() and re.match(r"^r\d+$", d.name)
+        )
 
     @model_validator(mode="after")
-    def _check_required_dirs(self) -> SubmissionDir:
-        for name in ("systems", "pareto"):
-            path = self.root / name
-            if path.is_dir():
+    def _check_sweep_files(self) -> ModelDir:
+        for fname in ("sweep_summary.csv", "sweep_distributions.csv"):
+            path = self.root / fname
+            if path.is_file():
                 self._check_results.append(
-                    ok("required-dir", f"Found required directory: {name}/", path, "#1")
+                    ok(
+                        "model-sweep-file",
+                        f"Found {fname} in {self.system_name}/{self.model_name}/",
+                        path,
+                        "#1",
+                    )
                 )
             else:
                 self._check_results.append(
-                    err("required-dir", f"Missing required directory: {name}/", path, "#1")
+                    err(
+                        "model-sweep-file",
+                        f"Missing {fname} in {self.system_name}/{self.model_name}/",
+                        path,
+                        "#1",
+                    )
                 )
         return self
 
 
-class SrcDir(BaseModel):
-    """Validates src/ exists for Standardized division submissions (§2.2.1)."""
+class RunDir(BaseModel):
+    """Validates an r<N>/ run directory: required files for a self-contained pareto point."""
 
     _check_results: list[CheckResult] = PrivateAttr(default_factory=list)
 
     root: Path
-    division: Division
+    system_name: str
+    model_name: str
 
     @model_validator(mode="after")
-    def _check_src(self) -> SrcDir:
-        if self.division != Division.STANDARDIZED:
-            return self
-        src_dir = self.root / "src"
-        if src_dir.is_dir():
+    def _check_run_files(self) -> RunDir:
+        label = f"{self.system_name}/{self.model_name}/{self.root.name}"
+
+        point_yaml = self.root / "point.yaml"
+        if point_yaml.is_file():
             self._check_results.append(
-                ok("src-dir", "src/ present (required for Standardized division)", src_dir, "#1")
+                ok("run-point-config", f"Found point.yaml in {label}/", point_yaml, "#1")
             )
         else:
             self._check_results.append(
-                err(
-                    "src-dir",
-                    "Missing src/ directory (required for Standardized division)",
-                    src_dir,
-                    "#1",
-                )
+                err("run-point-config", f"Missing point.yaml in {label}/", point_yaml, "#1")
             )
-        return self
-
-
-class SystemPareto(BaseModel):
-    """Validates pareto/<system_id>/ exists."""
-
-    _check_results: list[CheckResult] = PrivateAttr(default_factory=list)
-
-    pareto_dir: Path
-    system_id: str
-
-    @computed_field  # type: ignore[prop-decorator]
-    @property
-    def system_dir(self) -> Path:
-        return self.pareto_dir / self.system_id
 
-    @model_validator(mode="after")
-    def _check_dir_exists(self) -> SystemPareto:
-        path = self.pareto_dir / self.system_id
-        if path.is_dir():
+        accuracy_dir = self.root / "accuracy"
+        if accuracy_dir.is_dir():
             self._check_results.append(
-                ok("pareto-dir-exists", f"Found pareto/{self.system_id}/", path, "#1")
+                ok("run-accuracy-dir", f"Found accuracy/ in {label}/", accuracy_dir, "#1")
             )
         else:
             self._check_results.append(
-                err("pareto-dir-exists", f"No pareto/{self.system_id}/ directory found", path, "#1")
+                err("run-accuracy-dir", f"Missing accuracy/ in {label}/", accuracy_dir, "#1")
             )
-        return self
-
-
-class ModelDir(BaseModel):
-    """Validates points/, results/, and accuracy/ exist under a benchmark-model directory."""
-
-    _check_results: list[CheckResult] = PrivateAttr(default_factory=list)
-
-    root: Path
-    system_id: str
-    benchmark_model: str
-
-    @computed_field  # type: ignore[prop-decorator]
-    @property
-    def points_dir(self) -> Path:
-        return self.root / "points"
-
-    @computed_field  # type: ignore[prop-decorator]
-    @property
-    def results_dir(self) -> Path:
-        return self.root / "results"
-
-    @computed_field  # type: ignore[prop-decorator]
-    @property
-    def accuracy_dir(self) -> Path:
-        return self.root / "accuracy"
 
-    @model_validator(mode="after")
-    def _check_subdirs(self) -> ModelDir:
-        for name in ("points", "results", "accuracy"):
-            path = self.root / name
-            if path.is_dir():
+        for fname in ("mlperf_endpoints_log_summary.json", "mlperf_endpoints_log_detail.json"):
+            path = self.root / fname
+            if path.is_file():
                 self._check_results.append(
-                    ok(
-                        "pareto-subdir",
-                        f"Found {name}/ in pareto/{self.system_id}/{self.benchmark_model}/",
-                        path,
-                        "#1",
-                    )
+                    ok("run-result-files", f"Found {fname} in {label}/", path, "#1")
                 )
             else:
                 self._check_results.append(
-                    err(
-                        "pareto-subdir",
-                        f"Missing {name}/ in pareto/{self.system_id}/{self.benchmark_model}/",
-                        path,
-                        "#1",
-                    )
+                    err("run-result-files", f"Missing {fname} in {label}/", path, "#1")
+                )
+
+        for fname, rule in (("run_metadata.json", "run-metadata"), ("report.txt", "run-report")):
+            path = self.root / fname
+            if path.is_file():
+                self._check_results.append(ok(rule, f"Found {fname} in {label}/", path, "#1"))
+            else:
+                self._check_results.append(
+                    err(rule, f"Missing {fname} in {label}/", path, "#1")
                 )
+
+        src_dir = self.root / "src"
+        impl_dirs = [d for d in src_dir.iterdir() if d.is_dir()] if src_dir.is_dir() else []
+        if src_dir.is_dir() and impl_dirs:
+            self._check_results.append(
+                ok("run-src-dir", f"Found src/<impl>/ in {label}/", src_dir, "#1")
+            )
+        else:
+            self._check_results.append(
+                err(
+                    "run-src-dir",
+                    f"Missing src/<implementation>/ in {label}/ (src/ must exist with at least one implementation subdirectory)",
+                    src_dir,
+                    "#1",
+                )
+            )
+
         return self
diff --git a/test_submissions/invalid_submission/pareto/bad_system/llama3-70b/accuracy/accuracy.txt b/test_submissions/invalid_submission/bad_system/llama3-70b/r16/accuracy/accuracy.txt
similarity index 100%
rename from test_submissions/invalid_submission/pareto/bad_system/llama3-70b/accuracy/accuracy.txt
rename to test_submissions/invalid_submission/bad_system/llama3-70b/r16/accuracy/accuracy.txt
diff --git a/test_submissions/invalid_submission/pareto/bad_system/llama3-70b/accuracy/accuracy_result.json b/test_submissions/invalid_submission/bad_system/llama3-70b/r16/accuracy/accuracy_result.json
similarity index 100%
rename from test_submissions/invalid_submission/pareto/bad_system/llama3-70b/accuracy/accuracy_result.json
rename to test_submissions/invalid_submission/bad_system/llama3-70b/r16/accuracy/accuracy_result.json
diff --git a/test_submissions/invalid_submission/pareto/bad_system/llama3-70b/results/point_16/mlperf_endpoints_log_detail.json b/test_submissions/invalid_submission/bad_system/llama3-70b/r16/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/invalid_submission/pareto/bad_system/llama3-70b/results/point_16/mlperf_endpoints_log_detail.json
rename to test_submissions/invalid_submission/bad_system/llama3-70b/r16/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/invalid_submission/pareto/bad_system/llama3-70b/results/point_16/mlperf_endpoints_log_summary.json b/test_submissions/invalid_submission/bad_system/llama3-70b/r16/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/invalid_submission/pareto/bad_system/llama3-70b/results/point_16/mlperf_endpoints_log_summary.json
rename to test_submissions/invalid_submission/bad_system/llama3-70b/r16/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/invalid_submission/pareto/bad_system/llama3-70b/points/point_16.yaml b/test_submissions/invalid_submission/bad_system/llama3-70b/r16/point.yaml
similarity index 100%
rename from test_submissions/invalid_submission/pareto/bad_system/llama3-70b/points/point_16.yaml
rename to test_submissions/invalid_submission/bad_system/llama3-70b/r16/point.yaml
diff --git a/test_submissions/invalid_submission/bad_system/llama3-70b/r16/report.txt b/test_submissions/invalid_submission/bad_system/llama3-70b/r16/report.txt
new file mode 100644
index 0000000..7268543
--- /dev/null
+++ b/test_submissions/invalid_submission/bad_system/llama3-70b/r16/report.txt
@@ -0,0 +1 @@
+Run r16 — autogenerated report placeholder
diff --git a/test_submissions/invalid_submission/bad_system/llama3-70b/r16/run_metadata.json b/test_submissions/invalid_submission/bad_system/llama3-70b/r16/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/invalid_submission/bad_system/llama3-70b/r16/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/valid_standardized/src/.gitkeep b/test_submissions/invalid_submission/bad_system/llama3-70b/r16/src/vllm/.gitkeep
similarity index 100%
rename from test_submissions/valid_standardized/src/.gitkeep
rename to test_submissions/invalid_submission/bad_system/llama3-70b/r16/src/vllm/.gitkeep
diff --git a/test_submissions/invalid_submission/bad_system/llama3-70b/r38/accuracy/accuracy.txt b/test_submissions/invalid_submission/bad_system/llama3-70b/r38/accuracy/accuracy.txt
new file mode 100644
index 0000000..dc9bfd8
--- /dev/null
+++ b/test_submissions/invalid_submission/bad_system/llama3-70b/r38/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.38
diff --git a/test_submissions/invalid_submission/bad_system/llama3-70b/r38/accuracy/accuracy_result.json b/test_submissions/invalid_submission/bad_system/llama3-70b/r38/accuracy/accuracy_result.json
new file mode 100644
index 0000000..51689f2
--- /dev/null
+++ b/test_submissions/invalid_submission/bad_system/llama3-70b/r38/accuracy/accuracy_result.json
@@ -0,0 +1 @@
+{"metric": "rouge1", "score": 0.38, "quality_target": 0.43, "passed": false}
\ No newline at end of file
diff --git a/test_submissions/invalid_submission/pareto/bad_system/llama3-70b/results/point_38/mlperf_endpoints_log_detail.json b/test_submissions/invalid_submission/bad_system/llama3-70b/r38/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/invalid_submission/pareto/bad_system/llama3-70b/results/point_38/mlperf_endpoints_log_detail.json
rename to test_submissions/invalid_submission/bad_system/llama3-70b/r38/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/invalid_submission/pareto/bad_system/llama3-70b/results/point_38/mlperf_endpoints_log_summary.json b/test_submissions/invalid_submission/bad_system/llama3-70b/r38/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/invalid_submission/pareto/bad_system/llama3-70b/results/point_38/mlperf_endpoints_log_summary.json
rename to test_submissions/invalid_submission/bad_system/llama3-70b/r38/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/invalid_submission/pareto/bad_system/llama3-70b/points/point_38.yaml b/test_submissions/invalid_submission/bad_system/llama3-70b/r38/point.yaml
similarity index 100%
rename from test_submissions/invalid_submission/pareto/bad_system/llama3-70b/points/point_38.yaml
rename to test_submissions/invalid_submission/bad_system/llama3-70b/r38/point.yaml
diff --git a/test_submissions/invalid_submission/bad_system/llama3-70b/r38/report.txt b/test_submissions/invalid_submission/bad_system/llama3-70b/r38/report.txt
new file mode 100644
index 0000000..e151c43
--- /dev/null
+++ b/test_submissions/invalid_submission/bad_system/llama3-70b/r38/report.txt
@@ -0,0 +1 @@
+Run r38 — autogenerated report placeholder
diff --git a/test_submissions/invalid_submission/bad_system/llama3-70b/r38/run_metadata.json b/test_submissions/invalid_submission/bad_system/llama3-70b/r38/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/invalid_submission/bad_system/llama3-70b/r38/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/invalid_submission/bad_system/llama3-70b/r38/src/vllm/.gitkeep b/test_submissions/invalid_submission/bad_system/llama3-70b/r38/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/invalid_submission/bad_system/llama3-70b/r88/accuracy/accuracy.txt b/test_submissions/invalid_submission/bad_system/llama3-70b/r88/accuracy/accuracy.txt
new file mode 100644
index 0000000..dc9bfd8
--- /dev/null
+++ b/test_submissions/invalid_submission/bad_system/llama3-70b/r88/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.38
diff --git a/test_submissions/invalid_submission/bad_system/llama3-70b/r88/accuracy/accuracy_result.json b/test_submissions/invalid_submission/bad_system/llama3-70b/r88/accuracy/accuracy_result.json
new file mode 100644
index 0000000..51689f2
--- /dev/null
+++ b/test_submissions/invalid_submission/bad_system/llama3-70b/r88/accuracy/accuracy_result.json
@@ -0,0 +1 @@
+{"metric": "rouge1", "score": 0.38, "quality_target": 0.43, "passed": false}
\ No newline at end of file
diff --git a/test_submissions/invalid_submission/pareto/bad_system/llama3-70b/results/point_88/mlperf_endpoints_log_detail.json b/test_submissions/invalid_submission/bad_system/llama3-70b/r88/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/invalid_submission/pareto/bad_system/llama3-70b/results/point_88/mlperf_endpoints_log_detail.json
rename to test_submissions/invalid_submission/bad_system/llama3-70b/r88/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/invalid_submission/pareto/bad_system/llama3-70b/results/point_88/mlperf_endpoints_log_summary.json b/test_submissions/invalid_submission/bad_system/llama3-70b/r88/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/invalid_submission/pareto/bad_system/llama3-70b/results/point_88/mlperf_endpoints_log_summary.json
rename to test_submissions/invalid_submission/bad_system/llama3-70b/r88/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/invalid_submission/pareto/bad_system/llama3-70b/points/point_88.yaml b/test_submissions/invalid_submission/bad_system/llama3-70b/r88/point.yaml
similarity index 100%
rename from test_submissions/invalid_submission/pareto/bad_system/llama3-70b/points/point_88.yaml
rename to test_submissions/invalid_submission/bad_system/llama3-70b/r88/point.yaml
diff --git a/test_submissions/invalid_submission/bad_system/llama3-70b/r88/report.txt b/test_submissions/invalid_submission/bad_system/llama3-70b/r88/report.txt
new file mode 100644
index 0000000..cbb894a
--- /dev/null
+++ b/test_submissions/invalid_submission/bad_system/llama3-70b/r88/report.txt
@@ -0,0 +1 @@
+Run r88 — autogenerated report placeholder
diff --git a/test_submissions/invalid_submission/bad_system/llama3-70b/r88/run_metadata.json b/test_submissions/invalid_submission/bad_system/llama3-70b/r88/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/invalid_submission/bad_system/llama3-70b/r88/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/invalid_submission/bad_system/llama3-70b/r88/src/vllm/.gitkeep b/test_submissions/invalid_submission/bad_system/llama3-70b/r88/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/invalid_submission/bad_system/llama3-70b/sweep_distributions.csv b/test_submissions/invalid_submission/bad_system/llama3-70b/sweep_distributions.csv
new file mode 100644
index 0000000..f501c1f
--- /dev/null
+++ b/test_submissions/invalid_submission/bad_system/llama3-70b/sweep_distributions.csv
@@ -0,0 +1 @@
+concurrency,percentile,ttft_ms,tpot_ms,output_tokens
diff --git a/test_submissions/invalid_submission/bad_system/llama3-70b/sweep_summary.csv b/test_submissions/invalid_submission/bad_system/llama3-70b/sweep_summary.csv
new file mode 100644
index 0000000..3faf209
--- /dev/null
+++ b/test_submissions/invalid_submission/bad_system/llama3-70b/sweep_summary.csv
@@ -0,0 +1 @@
+concurrency,qps,ttft_p50_ms,ttft_p95_ms,tpot_p50_ms,system_tps
diff --git a/test_submissions/invalid_submission/systems/bad_system.json b/test_submissions/invalid_submission/bad_system/system_desc.json
similarity index 100%
rename from test_submissions/invalid_submission/systems/bad_system.json
rename to test_submissions/invalid_submission/bad_system/system_desc.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/accuracy/accuracy.txt b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/accuracy/accuracy.txt
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/accuracy/accuracy.txt
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/accuracy/accuracy.txt
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/accuracy/accuracy_result.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/accuracy/accuracy_result.json
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/accuracy/accuracy_result.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/accuracy/accuracy_result.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_1024/mlperf_endpoints_log_detail.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_1024/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_1024/mlperf_endpoints_log_summary.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_1024/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_1024.yaml b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/point.yaml
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_1024.yaml
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/point.yaml
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/report.txt b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/report.txt
new file mode 100644
index 0000000..ef6db2b
--- /dev/null
+++ b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/report.txt
@@ -0,0 +1 @@
+Run r1024 — autogenerated report placeholder
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/run_metadata.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/src/vllm/.gitkeep b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r1024/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/accuracy/accuracy.txt b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/accuracy/accuracy.txt
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/accuracy/accuracy.txt
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/accuracy/accuracy.txt
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/accuracy/accuracy_result.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/accuracy/accuracy_result.json
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/accuracy/accuracy_result.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/accuracy/accuracy_result.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_128/mlperf_endpoints_log_detail.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_128/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_128/mlperf_endpoints_log_summary.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_128/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_128.yaml b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/point.yaml
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_128.yaml
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/point.yaml
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/report.txt b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/report.txt
new file mode 100644
index 0000000..2128812
--- /dev/null
+++ b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/report.txt
@@ -0,0 +1 @@
+Run r128 — autogenerated report placeholder
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/run_metadata.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/src/vllm/.gitkeep b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r128/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/accuracy/accuracy.txt b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/accuracy/accuracy.txt
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/accuracy/accuracy.txt
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/accuracy/accuracy.txt
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/accuracy/accuracy_result.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/accuracy/accuracy_result.json
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/accuracy/accuracy_result.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/accuracy/accuracy_result.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_16/mlperf_endpoints_log_detail.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_16/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_16/mlperf_endpoints_log_summary.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_16/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_16.yaml b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/point.yaml
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_16.yaml
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/point.yaml
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/report.txt b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/report.txt
new file mode 100644
index 0000000..7268543
--- /dev/null
+++ b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/report.txt
@@ -0,0 +1 @@
+Run r16 — autogenerated report placeholder
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/run_metadata.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/src/vllm/.gitkeep b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r16/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/accuracy/accuracy.txt b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/accuracy/accuracy.txt
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/accuracy/accuracy.txt
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/accuracy/accuracy.txt
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/accuracy/accuracy_result.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/accuracy/accuracy_result.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/accuracy/accuracy_result.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/accuracy/accuracy_result.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_2048/mlperf_endpoints_log_detail.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_2048/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_2048/mlperf_endpoints_log_summary.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_2048/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_2048.yaml b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/point.yaml
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_2048.yaml
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/point.yaml
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/report.txt b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/report.txt
new file mode 100644
index 0000000..d8874f4
--- /dev/null
+++ b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/report.txt
@@ -0,0 +1 @@
+Run r2048 — autogenerated report placeholder
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/run_metadata.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/src/vllm/.gitkeep b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r2048/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/accuracy/accuracy.txt b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/accuracy/accuracy.txt
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/accuracy/accuracy.txt
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/accuracy/accuracy.txt
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/accuracy/accuracy_result.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/accuracy/accuracy_result.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/accuracy/accuracy_result.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/accuracy/accuracy_result.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_4/mlperf_endpoints_log_detail.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_4/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_4/mlperf_endpoints_log_summary.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_4/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_4.yaml b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/point.yaml
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_4.yaml
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/point.yaml
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/report.txt b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/report.txt
new file mode 100644
index 0000000..024a6e2
--- /dev/null
+++ b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/report.txt
@@ -0,0 +1 @@
+Run r4 — autogenerated report placeholder
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/run_metadata.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/src/vllm/.gitkeep b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r4/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/accuracy/accuracy.txt b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/accuracy/accuracy.txt
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/accuracy/accuracy.txt
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/accuracy/accuracy.txt
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/accuracy/accuracy_result.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/accuracy/accuracy_result.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/accuracy/accuracy_result.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/accuracy/accuracy_result.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_512/mlperf_endpoints_log_detail.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_512/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_512/mlperf_endpoints_log_summary.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_512/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_512.yaml b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/point.yaml
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_512.yaml
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/point.yaml
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/report.txt b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/report.txt
new file mode 100644
index 0000000..ea5355f
--- /dev/null
+++ b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/report.txt
@@ -0,0 +1 @@
+Run r512 — autogenerated report placeholder
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/run_metadata.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/src/vllm/.gitkeep b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r512/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/accuracy/accuracy.txt b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/accuracy/accuracy.txt
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/accuracy/accuracy.txt
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/accuracy/accuracy.txt
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/accuracy/accuracy_result.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/accuracy/accuracy_result.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/accuracy/accuracy_result.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/accuracy/accuracy_result.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_64/mlperf_endpoints_log_detail.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_64/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_64/mlperf_endpoints_log_summary.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/results/point_64/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_64.yaml b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/point.yaml
similarity index 100%
rename from test_submissions/sub_a/pareto/sys_mi355x_8gpu/gpt-oss-120b/points/point_64.yaml
rename to test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/point.yaml
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/report.txt b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/report.txt
new file mode 100644
index 0000000..76f16dd
--- /dev/null
+++ b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/report.txt
@@ -0,0 +1 @@
+Run r64 — autogenerated report placeholder
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/run_metadata.json b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/src/vllm/.gitkeep b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/r64/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/sweep_distributions.csv b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/sweep_distributions.csv
new file mode 100644
index 0000000..f501c1f
--- /dev/null
+++ b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/sweep_distributions.csv
@@ -0,0 +1 @@
+concurrency,percentile,ttft_ms,tpot_ms,output_tokens
diff --git a/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/sweep_summary.csv b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/sweep_summary.csv
new file mode 100644
index 0000000..3faf209
--- /dev/null
+++ b/test_submissions/sub_a/sys_mi355x_8gpu/gpt-oss-120b/sweep_summary.csv
@@ -0,0 +1 @@
+concurrency,qps,ttft_p50_ms,ttft_p95_ms,tpot_p50_ms,system_tps
diff --git a/test_submissions/sub_a/systems/sys_mi355x_8gpu.json b/test_submissions/sub_a/sys_mi355x_8gpu/system_desc.json
similarity index 100%
rename from test_submissions/sub_a/systems/sys_mi355x_8gpu.json
rename to test_submissions/sub_a/sys_mi355x_8gpu/system_desc.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/accuracy/accuracy.txt b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/accuracy/accuracy.txt
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/accuracy/accuracy.txt
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/accuracy/accuracy.txt
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/accuracy/accuracy_result.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/accuracy/accuracy_result.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/accuracy/accuracy_result.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/accuracy/accuracy_result.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_1024/mlperf_endpoints_log_detail.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_1024/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_1024/mlperf_endpoints_log_summary.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_1024/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_1024.yaml b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/point.yaml
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_1024.yaml
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/point.yaml
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/report.txt b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/report.txt
new file mode 100644
index 0000000..ef6db2b
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/report.txt
@@ -0,0 +1 @@
+Run r1024 — autogenerated report placeholder
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/run_metadata.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/src/vllm/.gitkeep b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r1024/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/accuracy/accuracy.txt b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/accuracy/accuracy.txt
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/accuracy/accuracy.txt
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/accuracy/accuracy.txt
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/accuracy/accuracy_result.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/accuracy/accuracy_result.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/accuracy/accuracy_result.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/accuracy/accuracy_result.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_128/mlperf_endpoints_log_detail.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_128/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_128/mlperf_endpoints_log_summary.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_128/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_128.yaml b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/point.yaml
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_128.yaml
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/point.yaml
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/report.txt b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/report.txt
new file mode 100644
index 0000000..2128812
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/report.txt
@@ -0,0 +1 @@
+Run r128 — autogenerated report placeholder
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/run_metadata.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/src/vllm/.gitkeep b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r128/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/accuracy/accuracy.txt b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/accuracy/accuracy.txt
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/accuracy/accuracy.txt
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/accuracy/accuracy.txt
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/accuracy/accuracy_result.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/accuracy/accuracy_result.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/accuracy/accuracy_result.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/accuracy/accuracy_result.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_16/mlperf_endpoints_log_detail.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_16/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_16/mlperf_endpoints_log_summary.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_16/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_16.yaml b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/point.yaml
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_16.yaml
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/point.yaml
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/report.txt b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/report.txt
new file mode 100644
index 0000000..7268543
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/report.txt
@@ -0,0 +1 @@
+Run r16 — autogenerated report placeholder
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/run_metadata.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/src/vllm/.gitkeep b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r16/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/accuracy/accuracy.txt b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/accuracy/accuracy_result.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_2048/mlperf_endpoints_log_detail.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_2048/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_2048/mlperf_endpoints_log_summary.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_2048/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_2048.yaml b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/point.yaml
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_2048.yaml
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/point.yaml
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/report.txt b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/report.txt
new file mode 100644
index 0000000..d8874f4
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/report.txt
@@ -0,0 +1 @@
+Run r2048 — autogenerated report placeholder
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/run_metadata.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/src/vllm/.gitkeep b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r2048/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/accuracy/accuracy.txt b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/accuracy/accuracy_result.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_4/mlperf_endpoints_log_detail.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_4/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_4/mlperf_endpoints_log_summary.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_4/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_4.yaml b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/point.yaml
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_4.yaml
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/point.yaml
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/report.txt b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/report.txt
new file mode 100644
index 0000000..024a6e2
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/report.txt
@@ -0,0 +1 @@
+Run r4 — autogenerated report placeholder
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/run_metadata.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/src/vllm/.gitkeep b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r4/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/accuracy/accuracy.txt b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/accuracy/accuracy_result.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_512/mlperf_endpoints_log_detail.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_512/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_512/mlperf_endpoints_log_summary.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_512/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_512.yaml b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/point.yaml
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_512.yaml
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/point.yaml
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/report.txt b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/report.txt
new file mode 100644
index 0000000..ea5355f
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/report.txt
@@ -0,0 +1 @@
+Run r512 — autogenerated report placeholder
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/run_metadata.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/src/vllm/.gitkeep b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r512/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/accuracy/accuracy.txt b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/accuracy/accuracy_result.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_64/mlperf_endpoints_log_detail.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_64/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_64/mlperf_endpoints_log_summary.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/results/point_64/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_64.yaml b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/point.yaml
similarity index 100%
rename from test_submissions/sub_b/pareto/sys_mi355x_16gpu/gpt-oss-120b/points/point_64.yaml
rename to test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/point.yaml
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/report.txt b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/report.txt
new file mode 100644
index 0000000..76f16dd
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/report.txt
@@ -0,0 +1 @@
+Run r64 — autogenerated report placeholder
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/run_metadata.json b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/src/vllm/.gitkeep b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/r64/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/sweep_distributions.csv b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/sweep_distributions.csv
new file mode 100644
index 0000000..f501c1f
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/sweep_distributions.csv
@@ -0,0 +1 @@
+concurrency,percentile,ttft_ms,tpot_ms,output_tokens
diff --git a/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/sweep_summary.csv b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/sweep_summary.csv
new file mode 100644
index 0000000..3faf209
--- /dev/null
+++ b/test_submissions/sub_b/sys_mi355x_16gpu/gpt-oss-120b/sweep_summary.csv
@@ -0,0 +1 @@
+concurrency,qps,ttft_p50_ms,ttft_p95_ms,tpot_p50_ms,system_tps
diff --git a/test_submissions/sub_b/systems/sys_mi355x_16gpu.json b/test_submissions/sub_b/sys_mi355x_16gpu/system_desc.json
similarity index 100%
rename from test_submissions/sub_b/systems/sys_mi355x_16gpu.json
rename to test_submissions/sub_b/sys_mi355x_16gpu/system_desc.json
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/accuracy/accuracy.txt b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/accuracy/accuracy_result.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_128/mlperf_endpoints_log_detail.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_128/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_128/mlperf_endpoints_log_summary.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_128/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_128.yaml b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/point.yaml
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_128.yaml
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/point.yaml
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/report.txt b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/report.txt
new file mode 100644
index 0000000..2128812
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/report.txt
@@ -0,0 +1 @@
+Run r128 — autogenerated report placeholder
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/run_metadata.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/src/vllm/.gitkeep b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r128/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/accuracy/accuracy.txt b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/accuracy/accuracy_result.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_16/mlperf_endpoints_log_detail.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_16/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_16/mlperf_endpoints_log_summary.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_16/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_16.yaml b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/point.yaml
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_16.yaml
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/point.yaml
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/report.txt b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/report.txt
new file mode 100644
index 0000000..7268543
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/report.txt
@@ -0,0 +1 @@
+Run r16 — autogenerated report placeholder
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/run_metadata.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/src/vllm/.gitkeep b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r16/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/accuracy/accuracy.txt b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/accuracy/accuracy_result.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_256/mlperf_endpoints_log_detail.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_256/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_256/mlperf_endpoints_log_summary.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_256/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_256.yaml b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/point.yaml
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_256.yaml
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/point.yaml
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/report.txt b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/report.txt
new file mode 100644
index 0000000..91b6b15
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/report.txt
@@ -0,0 +1 @@
+Run r256 — autogenerated report placeholder
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/run_metadata.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/src/vllm/.gitkeep b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r256/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/accuracy/accuracy.txt b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/accuracy/accuracy_result.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_32/mlperf_endpoints_log_detail.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_32/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_32/mlperf_endpoints_log_summary.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_32/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_32.yaml b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/point.yaml
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_32.yaml
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/point.yaml
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/report.txt b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/report.txt
new file mode 100644
index 0000000..51064e6
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/report.txt
@@ -0,0 +1 @@
+Run r32 — autogenerated report placeholder
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/run_metadata.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/src/vllm/.gitkeep b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r32/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/accuracy/accuracy.txt b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/accuracy/accuracy_result.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_512/mlperf_endpoints_log_detail.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_512/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_512/mlperf_endpoints_log_summary.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_512/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_512.yaml b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/point.yaml
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_512.yaml
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/point.yaml
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/report.txt b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/report.txt
new file mode 100644
index 0000000..ea5355f
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/report.txt
@@ -0,0 +1 @@
+Run r512 — autogenerated report placeholder
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/run_metadata.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/src/vllm/.gitkeep b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r512/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/accuracy/accuracy.txt b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/accuracy/accuracy_result.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_64/mlperf_endpoints_log_detail.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_64/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_64/mlperf_endpoints_log_summary.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_64/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_64.yaml b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/point.yaml
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_64.yaml
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/point.yaml
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/report.txt b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/report.txt
new file mode 100644
index 0000000..76f16dd
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/report.txt
@@ -0,0 +1 @@
+Run r64 — autogenerated report placeholder
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/run_metadata.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/src/vllm/.gitkeep b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r64/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/accuracy/accuracy.txt b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/accuracy/accuracy_result.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_8/mlperf_endpoints_log_detail.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_8/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_8/mlperf_endpoints_log_summary.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/results/point_8/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_8.yaml b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/point.yaml
similarity index 100%
rename from test_submissions/sub_c/pareto/sys_tpu_4chip/qwen3-coder-480b/points/point_8.yaml
rename to test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/point.yaml
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/report.txt b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/report.txt
new file mode 100644
index 0000000..e4c38ca
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/report.txt
@@ -0,0 +1 @@
+Run r8 — autogenerated report placeholder
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/run_metadata.json b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/src/vllm/.gitkeep b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/r8/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/sweep_distributions.csv b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/sweep_distributions.csv
new file mode 100644
index 0000000..f501c1f
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/sweep_distributions.csv
@@ -0,0 +1 @@
+concurrency,percentile,ttft_ms,tpot_ms,output_tokens
diff --git a/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/sweep_summary.csv b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/sweep_summary.csv
new file mode 100644
index 0000000..3faf209
--- /dev/null
+++ b/test_submissions/sub_c/sys_tpu_4chip/qwen3-coder-480b/sweep_summary.csv
@@ -0,0 +1 @@
+concurrency,qps,ttft_p50_ms,ttft_p95_ms,tpot_p50_ms,system_tps
diff --git a/test_submissions/sub_c/systems/sys_tpu_4chip.json b/test_submissions/sub_c/sys_tpu_4chip/system_desc.json
similarity index 100%
rename from test_submissions/sub_c/systems/sys_tpu_4chip.json
rename to test_submissions/sub_c/sys_tpu_4chip/system_desc.json
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/accuracy/accuracy.txt b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/accuracy/accuracy_result.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_1024/mlperf_endpoints_log_detail.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_1024/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_1024/mlperf_endpoints_log_summary.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_1024/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_1024.yaml b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/point.yaml
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_1024.yaml
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/point.yaml
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/report.txt b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/report.txt
new file mode 100644
index 0000000..ef6db2b
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/report.txt
@@ -0,0 +1 @@
+Run r1024 — autogenerated report placeholder
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/run_metadata.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/src/vllm/.gitkeep b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r1024/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/accuracy/accuracy.txt b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/accuracy/accuracy_result.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_128/mlperf_endpoints_log_detail.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_128/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_128/mlperf_endpoints_log_summary.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_128/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_128.yaml b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/point.yaml
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_128.yaml
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/point.yaml
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/report.txt b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/report.txt
new file mode 100644
index 0000000..2128812
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/report.txt
@@ -0,0 +1 @@
+Run r128 — autogenerated report placeholder
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/run_metadata.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/src/vllm/.gitkeep b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r128/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/accuracy/accuracy.txt b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/accuracy/accuracy_result.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_16/mlperf_endpoints_log_detail.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_16/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_16/mlperf_endpoints_log_summary.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_16/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_16.yaml b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/point.yaml
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_16.yaml
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/point.yaml
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/report.txt b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/report.txt
new file mode 100644
index 0000000..7268543
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/report.txt
@@ -0,0 +1 @@
+Run r16 — autogenerated report placeholder
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/run_metadata.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/src/vllm/.gitkeep b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r16/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/accuracy/accuracy.txt b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/accuracy/accuracy_result.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_256/mlperf_endpoints_log_detail.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_256/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_256/mlperf_endpoints_log_summary.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_256/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_256.yaml b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/point.yaml
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_256.yaml
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/point.yaml
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/report.txt b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/report.txt
new file mode 100644
index 0000000..91b6b15
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/report.txt
@@ -0,0 +1 @@
+Run r256 — autogenerated report placeholder
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/run_metadata.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/src/vllm/.gitkeep b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r256/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/accuracy/accuracy.txt b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/accuracy/accuracy_result.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_32/mlperf_endpoints_log_detail.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_32/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_32/mlperf_endpoints_log_summary.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_32/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_32.yaml b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/point.yaml
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_32.yaml
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/point.yaml
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/report.txt b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/report.txt
new file mode 100644
index 0000000..51064e6
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/report.txt
@@ -0,0 +1 @@
+Run r32 — autogenerated report placeholder
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/run_metadata.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/src/vllm/.gitkeep b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r32/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/accuracy/accuracy.txt b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/accuracy/accuracy_result.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_512/mlperf_endpoints_log_detail.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_512/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_512/mlperf_endpoints_log_summary.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_512/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_512.yaml b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/point.yaml
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_512.yaml
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/point.yaml
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/report.txt b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/report.txt
new file mode 100644
index 0000000..ea5355f
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/report.txt
@@ -0,0 +1 @@
+Run r512 — autogenerated report placeholder
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/run_metadata.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/src/vllm/.gitkeep b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r512/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/accuracy/accuracy.txt b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/accuracy/accuracy_result.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_64/mlperf_endpoints_log_detail.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_64/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_64/mlperf_endpoints_log_summary.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_64/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_64.yaml b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/point.yaml
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_64.yaml
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/point.yaml
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/report.txt b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/report.txt
new file mode 100644
index 0000000..76f16dd
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/report.txt
@@ -0,0 +1 @@
+Run r64 — autogenerated report placeholder
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/run_metadata.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/src/vllm/.gitkeep b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r64/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/accuracy/accuracy.txt b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/accuracy/accuracy_result.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_8/mlperf_endpoints_log_detail.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_8/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_8/mlperf_endpoints_log_summary.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/results/point_8/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_8.yaml b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/point.yaml
similarity index 100%
rename from test_submissions/sub_d/pareto/sys_tpu_8chip/qwen3-coder-480b/points/point_8.yaml
rename to test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/point.yaml
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/report.txt b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/report.txt
new file mode 100644
index 0000000..e4c38ca
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/report.txt
@@ -0,0 +1 @@
+Run r8 — autogenerated report placeholder
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/run_metadata.json b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/src/vllm/.gitkeep b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/r8/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/sweep_distributions.csv b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/sweep_distributions.csv
new file mode 100644
index 0000000..f501c1f
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/sweep_distributions.csv
@@ -0,0 +1 @@
+concurrency,percentile,ttft_ms,tpot_ms,output_tokens
diff --git a/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/sweep_summary.csv b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/sweep_summary.csv
new file mode 100644
index 0000000..3faf209
--- /dev/null
+++ b/test_submissions/sub_d/sys_tpu_8chip/qwen3-coder-480b/sweep_summary.csv
@@ -0,0 +1 @@
+concurrency,qps,ttft_p50_ms,ttft_p95_ms,tpot_p50_ms,system_tps
diff --git a/test_submissions/sub_d/systems/sys_tpu_8chip.json b/test_submissions/sub_d/sys_tpu_8chip/system_desc.json
similarity index 100%
rename from test_submissions/sub_d/systems/sys_tpu_8chip.json
rename to test_submissions/sub_d/sys_tpu_8chip/system_desc.json
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/accuracy/accuracy.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/accuracy/accuracy_result.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_1/mlperf_endpoints_log_detail.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_1/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_1/mlperf_endpoints_log_summary.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_1/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_1.yaml b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/point.yaml
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_1.yaml
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/point.yaml
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/report.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/report.txt
new file mode 100644
index 0000000..bf18a4f
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/report.txt
@@ -0,0 +1 @@
+Run r1 — autogenerated report placeholder
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/run_metadata.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/src/vllm/.gitkeep b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/accuracy/accuracy.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/accuracy/accuracy_result.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_1024/mlperf_endpoints_log_detail.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_1024/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_1024/mlperf_endpoints_log_summary.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_1024/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_1024.yaml b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/point.yaml
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_1024.yaml
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/point.yaml
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/report.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/report.txt
new file mode 100644
index 0000000..ef6db2b
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/report.txt
@@ -0,0 +1 @@
+Run r1024 — autogenerated report placeholder
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/run_metadata.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/src/vllm/.gitkeep b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r1024/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/accuracy/accuracy.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/accuracy/accuracy_result.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_128/mlperf_endpoints_log_detail.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_128/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_128/mlperf_endpoints_log_summary.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_128/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_128.yaml b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/point.yaml
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_128.yaml
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/point.yaml
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/report.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/report.txt
new file mode 100644
index 0000000..2128812
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/report.txt
@@ -0,0 +1 @@
+Run r128 — autogenerated report placeholder
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/run_metadata.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/src/vllm/.gitkeep b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r128/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/accuracy/accuracy.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/accuracy/accuracy_result.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_16/mlperf_endpoints_log_detail.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_16/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_16/mlperf_endpoints_log_summary.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_16/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_16.yaml b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/point.yaml
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_16.yaml
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/point.yaml
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/report.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/report.txt
new file mode 100644
index 0000000..7268543
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/report.txt
@@ -0,0 +1 @@
+Run r16 — autogenerated report placeholder
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/run_metadata.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/src/vllm/.gitkeep b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r16/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/accuracy/accuracy.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/accuracy/accuracy_result.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_2/mlperf_endpoints_log_detail.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_2/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_2/mlperf_endpoints_log_summary.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_2/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_2.yaml b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/point.yaml
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_2.yaml
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/point.yaml
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/report.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/report.txt
new file mode 100644
index 0000000..a07a826
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/report.txt
@@ -0,0 +1 @@
+Run r2 — autogenerated report placeholder
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/run_metadata.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/src/vllm/.gitkeep b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r2/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/accuracy/accuracy.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/accuracy/accuracy_result.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_256/mlperf_endpoints_log_detail.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_256/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_256/mlperf_endpoints_log_summary.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_256/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_256.yaml b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/point.yaml
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_256.yaml
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/point.yaml
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/report.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/report.txt
new file mode 100644
index 0000000..91b6b15
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/report.txt
@@ -0,0 +1 @@
+Run r256 — autogenerated report placeholder
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/run_metadata.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/src/vllm/.gitkeep b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r256/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/accuracy/accuracy.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/accuracy/accuracy_result.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_32/mlperf_endpoints_log_detail.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_32/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_32/mlperf_endpoints_log_summary.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_32/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_32.yaml b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/point.yaml
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_32.yaml
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/point.yaml
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/report.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/report.txt
new file mode 100644
index 0000000..51064e6
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/report.txt
@@ -0,0 +1 @@
+Run r32 — autogenerated report placeholder
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/run_metadata.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/src/vllm/.gitkeep b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r32/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/accuracy/accuracy.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/accuracy/accuracy_result.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_4/mlperf_endpoints_log_detail.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_4/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_4/mlperf_endpoints_log_summary.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_4/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_4.yaml b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/point.yaml
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_4.yaml
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/point.yaml
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/report.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/report.txt
new file mode 100644
index 0000000..024a6e2
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/report.txt
@@ -0,0 +1 @@
+Run r4 — autogenerated report placeholder
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/run_metadata.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/src/vllm/.gitkeep b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r4/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/accuracy/accuracy.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/accuracy/accuracy_result.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_512/mlperf_endpoints_log_detail.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_512/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_512/mlperf_endpoints_log_summary.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_512/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_512.yaml b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/point.yaml
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_512.yaml
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/point.yaml
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/report.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/report.txt
new file mode 100644
index 0000000..ea5355f
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/report.txt
@@ -0,0 +1 @@
+Run r512 — autogenerated report placeholder
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/run_metadata.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/src/vllm/.gitkeep b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r512/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/accuracy/accuracy.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/accuracy/accuracy_result.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_64/mlperf_endpoints_log_detail.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_64/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_64/mlperf_endpoints_log_summary.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_64/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_64.yaml b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/point.yaml
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_64.yaml
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/point.yaml
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/report.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/report.txt
new file mode 100644
index 0000000..76f16dd
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/report.txt
@@ -0,0 +1 @@
+Run r64 — autogenerated report placeholder
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/run_metadata.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/src/vllm/.gitkeep b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r64/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/accuracy/accuracy.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/accuracy/accuracy_result.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_8/mlperf_endpoints_log_detail.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_8/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_8/mlperf_endpoints_log_summary.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/results/point_8/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_8.yaml b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/point.yaml
similarity index 100%
rename from test_submissions/sub_e/pareto/sys_gaudi_dp1/llama3-8b/points/point_8.yaml
rename to test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/point.yaml
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/report.txt b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/report.txt
new file mode 100644
index 0000000..e4c38ca
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/report.txt
@@ -0,0 +1 @@
+Run r8 — autogenerated report placeholder
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/run_metadata.json b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/src/vllm/.gitkeep b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/r8/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/sweep_distributions.csv b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/sweep_distributions.csv
new file mode 100644
index 0000000..f501c1f
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/sweep_distributions.csv
@@ -0,0 +1 @@
+concurrency,percentile,ttft_ms,tpot_ms,output_tokens
diff --git a/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/sweep_summary.csv b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/sweep_summary.csv
new file mode 100644
index 0000000..3faf209
--- /dev/null
+++ b/test_submissions/sub_e/sys_gaudi_dp1/llama3-8b/sweep_summary.csv
@@ -0,0 +1 @@
+concurrency,qps,ttft_p50_ms,ttft_p95_ms,tpot_p50_ms,system_tps
diff --git a/test_submissions/sub_e/systems/sys_gaudi_dp1.json b/test_submissions/sub_e/sys_gaudi_dp1/system_desc.json
similarity index 100%
rename from test_submissions/sub_e/systems/sys_gaudi_dp1.json
rename to test_submissions/sub_e/sys_gaudi_dp1/system_desc.json
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/accuracy/accuracy.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/accuracy/accuracy_result.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_1/mlperf_endpoints_log_detail.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_1/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_1/mlperf_endpoints_log_summary.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_1/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_1.yaml b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/point.yaml
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_1.yaml
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/point.yaml
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/report.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/report.txt
new file mode 100644
index 0000000..bf18a4f
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/report.txt
@@ -0,0 +1 @@
+Run r1 — autogenerated report placeholder
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/run_metadata.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/src/vllm/.gitkeep b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/accuracy/accuracy.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/accuracy/accuracy_result.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_1024/mlperf_endpoints_log_detail.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_1024/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_1024/mlperf_endpoints_log_summary.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_1024/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_1024.yaml b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/point.yaml
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_1024.yaml
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/point.yaml
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/report.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/report.txt
new file mode 100644
index 0000000..ef6db2b
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/report.txt
@@ -0,0 +1 @@
+Run r1024 — autogenerated report placeholder
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/run_metadata.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/src/vllm/.gitkeep b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r1024/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/accuracy/accuracy.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/accuracy/accuracy_result.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_128/mlperf_endpoints_log_detail.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_128/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_128/mlperf_endpoints_log_summary.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_128/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_128.yaml b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/point.yaml
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_128.yaml
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/point.yaml
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/report.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/report.txt
new file mode 100644
index 0000000..2128812
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/report.txt
@@ -0,0 +1 @@
+Run r128 — autogenerated report placeholder
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/run_metadata.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/src/vllm/.gitkeep b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r128/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/accuracy/accuracy.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/accuracy/accuracy_result.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_16/mlperf_endpoints_log_detail.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_16/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_16/mlperf_endpoints_log_summary.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_16/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_16.yaml b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/point.yaml
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_16.yaml
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/point.yaml
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/report.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/report.txt
new file mode 100644
index 0000000..7268543
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/report.txt
@@ -0,0 +1 @@
+Run r16 — autogenerated report placeholder
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/run_metadata.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/src/vllm/.gitkeep b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r16/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/accuracy/accuracy.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/accuracy/accuracy_result.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_2/mlperf_endpoints_log_detail.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_2/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_2/mlperf_endpoints_log_summary.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_2/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_2.yaml b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/point.yaml
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_2.yaml
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/point.yaml
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/report.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/report.txt
new file mode 100644
index 0000000..a07a826
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/report.txt
@@ -0,0 +1 @@
+Run r2 — autogenerated report placeholder
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/run_metadata.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/src/vllm/.gitkeep b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r2/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/accuracy/accuracy.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/accuracy/accuracy_result.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_256/mlperf_endpoints_log_detail.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_256/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_256/mlperf_endpoints_log_summary.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_256/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_256.yaml b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/point.yaml
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_256.yaml
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/point.yaml
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/report.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/report.txt
new file mode 100644
index 0000000..91b6b15
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/report.txt
@@ -0,0 +1 @@
+Run r256 — autogenerated report placeholder
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/run_metadata.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/src/vllm/.gitkeep b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r256/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/accuracy/accuracy.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/accuracy/accuracy_result.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_32/mlperf_endpoints_log_detail.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_32/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_32/mlperf_endpoints_log_summary.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_32/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_32.yaml b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/point.yaml
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_32.yaml
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/point.yaml
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/report.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/report.txt
new file mode 100644
index 0000000..51064e6
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/report.txt
@@ -0,0 +1 @@
+Run r32 — autogenerated report placeholder
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/run_metadata.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/src/vllm/.gitkeep b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r32/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/accuracy/accuracy.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/accuracy/accuracy_result.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_4/mlperf_endpoints_log_detail.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_4/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_4/mlperf_endpoints_log_summary.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_4/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_4.yaml b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/point.yaml
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_4.yaml
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/point.yaml
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/report.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/report.txt
new file mode 100644
index 0000000..024a6e2
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/report.txt
@@ -0,0 +1 @@
+Run r4 — autogenerated report placeholder
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/run_metadata.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/src/vllm/.gitkeep b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r4/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/accuracy/accuracy.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/accuracy/accuracy_result.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_512/mlperf_endpoints_log_detail.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_512/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_512/mlperf_endpoints_log_summary.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_512/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_512.yaml b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/point.yaml
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_512.yaml
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/point.yaml
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/report.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/report.txt
new file mode 100644
index 0000000..ea5355f
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/report.txt
@@ -0,0 +1 @@
+Run r512 — autogenerated report placeholder
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/run_metadata.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/src/vllm/.gitkeep b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r512/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/accuracy/accuracy.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/accuracy/accuracy_result.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_64/mlperf_endpoints_log_detail.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_64/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_64/mlperf_endpoints_log_summary.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_64/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_64.yaml b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/point.yaml
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_64.yaml
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/point.yaml
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/report.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/report.txt
new file mode 100644
index 0000000..76f16dd
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/report.txt
@@ -0,0 +1 @@
+Run r64 — autogenerated report placeholder
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/run_metadata.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/src/vllm/.gitkeep b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r64/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/accuracy/accuracy.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/accuracy/accuracy_result.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_8/mlperf_endpoints_log_detail.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_8/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_8/mlperf_endpoints_log_summary.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/results/point_8/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_8.yaml b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/point.yaml
similarity index 100%
rename from test_submissions/sub_f/pareto/sys_gaudi_dp2/llama3-8b/points/point_8.yaml
rename to test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/point.yaml
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/report.txt b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/report.txt
new file mode 100644
index 0000000..e4c38ca
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/report.txt
@@ -0,0 +1 @@
+Run r8 — autogenerated report placeholder
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/run_metadata.json b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/src/vllm/.gitkeep b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/r8/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/sweep_distributions.csv b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/sweep_distributions.csv
new file mode 100644
index 0000000..f501c1f
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/sweep_distributions.csv
@@ -0,0 +1 @@
+concurrency,percentile,ttft_ms,tpot_ms,output_tokens
diff --git a/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/sweep_summary.csv b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/sweep_summary.csv
new file mode 100644
index 0000000..3faf209
--- /dev/null
+++ b/test_submissions/sub_f/sys_gaudi_dp2/llama3-8b/sweep_summary.csv
@@ -0,0 +1 @@
+concurrency,qps,ttft_p50_ms,ttft_p95_ms,tpot_p50_ms,system_tps
diff --git a/test_submissions/sub_f/systems/sys_gaudi_dp2.json b/test_submissions/sub_f/sys_gaudi_dp2/system_desc.json
similarity index 100%
rename from test_submissions/sub_f/systems/sys_gaudi_dp2.json
rename to test_submissions/sub_f/sys_gaudi_dp2/system_desc.json
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/accuracy/accuracy.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/accuracy/accuracy_result.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_1024/mlperf_endpoints_log_detail.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_1024/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_1024/mlperf_endpoints_log_summary.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_1024/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_1024.yaml b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/point.yaml
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_1024.yaml
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/point.yaml
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/report.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/report.txt
new file mode 100644
index 0000000..ef6db2b
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/report.txt
@@ -0,0 +1 @@
+Run r1024 — autogenerated report placeholder
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/run_metadata.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/src/vllm/.gitkeep b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1024/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/accuracy/accuracy.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/accuracy/accuracy_result.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_128/mlperf_endpoints_log_detail.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_128/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_128/mlperf_endpoints_log_summary.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_128/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_128.yaml b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/point.yaml
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_128.yaml
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/point.yaml
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/report.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/report.txt
new file mode 100644
index 0000000..2128812
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/report.txt
@@ -0,0 +1 @@
+Run r128 — autogenerated report placeholder
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/run_metadata.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/src/vllm/.gitkeep b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r128/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/accuracy/accuracy.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/accuracy/accuracy_result.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_1536/mlperf_endpoints_log_detail.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_1536/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_1536/mlperf_endpoints_log_summary.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_1536/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_1536.yaml b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/point.yaml
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_1536.yaml
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/point.yaml
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/report.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/report.txt
new file mode 100644
index 0000000..21dc3b9
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/report.txt
@@ -0,0 +1 @@
+Run r1536 — autogenerated report placeholder
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/run_metadata.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/src/vllm/.gitkeep b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r1536/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/accuracy/accuracy.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/accuracy/accuracy_result.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_192/mlperf_endpoints_log_detail.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_192/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_192/mlperf_endpoints_log_summary.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_192/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_192.yaml b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/point.yaml
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_192.yaml
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/point.yaml
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/report.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/report.txt
new file mode 100644
index 0000000..d6d1fdb
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/report.txt
@@ -0,0 +1 @@
+Run r192 — autogenerated report placeholder
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/run_metadata.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/src/vllm/.gitkeep b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r192/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/accuracy/accuracy.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/accuracy/accuracy_result.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_2048/mlperf_endpoints_log_detail.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_2048/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_2048/mlperf_endpoints_log_summary.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_2048/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_2048.yaml b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/point.yaml
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_2048.yaml
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/point.yaml
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/report.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/report.txt
new file mode 100644
index 0000000..d8874f4
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/report.txt
@@ -0,0 +1 @@
+Run r2048 — autogenerated report placeholder
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/run_metadata.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/src/vllm/.gitkeep b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r2048/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/accuracy/accuracy.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/accuracy/accuracy_result.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_256/mlperf_endpoints_log_detail.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_256/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_256/mlperf_endpoints_log_summary.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_256/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_256.yaml b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/point.yaml
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_256.yaml
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/point.yaml
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/report.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/report.txt
new file mode 100644
index 0000000..91b6b15
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/report.txt
@@ -0,0 +1 @@
+Run r256 — autogenerated report placeholder
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/run_metadata.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/src/vllm/.gitkeep b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r256/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/accuracy/accuracy.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/accuracy/accuracy_result.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_384/mlperf_endpoints_log_detail.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_384/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_384/mlperf_endpoints_log_summary.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_384/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_384.yaml b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/point.yaml
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_384.yaml
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/point.yaml
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/report.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/report.txt
new file mode 100644
index 0000000..95d6b85
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/report.txt
@@ -0,0 +1 @@
+Run r384 — autogenerated report placeholder
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/run_metadata.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/src/vllm/.gitkeep b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r384/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/accuracy/accuracy.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/accuracy/accuracy_result.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_512/mlperf_endpoints_log_detail.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_512/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_512/mlperf_endpoints_log_summary.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_512/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_512.yaml b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/point.yaml
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_512.yaml
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/point.yaml
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/report.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/report.txt
new file mode 100644
index 0000000..ea5355f
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/report.txt
@@ -0,0 +1 @@
+Run r512 — autogenerated report placeholder
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/run_metadata.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/src/vllm/.gitkeep b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r512/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/accuracy/accuracy.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/accuracy/accuracy_result.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_64/mlperf_endpoints_log_detail.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_64/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_64/mlperf_endpoints_log_summary.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_64/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_64.yaml b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/point.yaml
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_64.yaml
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/point.yaml
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/report.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/report.txt
new file mode 100644
index 0000000..76f16dd
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/report.txt
@@ -0,0 +1 @@
+Run r64 — autogenerated report placeholder
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/run_metadata.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/src/vllm/.gitkeep b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r64/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/accuracy/accuracy.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/accuracy/accuracy_result.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_768/mlperf_endpoints_log_detail.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_768/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_768/mlperf_endpoints_log_summary.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/results/point_768/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_768.yaml b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/point.yaml
similarity index 100%
rename from test_submissions/sub_g/pareto/sys_8gpu_vllm/llama3-70b/points/point_768.yaml
rename to test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/point.yaml
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/report.txt b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/report.txt
new file mode 100644
index 0000000..80814c9
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/report.txt
@@ -0,0 +1 @@
+Run r768 — autogenerated report placeholder
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/run_metadata.json b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/src/vllm/.gitkeep b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/r768/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/sweep_distributions.csv b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/sweep_distributions.csv
new file mode 100644
index 0000000..f501c1f
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/sweep_distributions.csv
@@ -0,0 +1 @@
+concurrency,percentile,ttft_ms,tpot_ms,output_tokens
diff --git a/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/sweep_summary.csv b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/sweep_summary.csv
new file mode 100644
index 0000000..3faf209
--- /dev/null
+++ b/test_submissions/sub_g/sys_8gpu_vllm/llama3-70b/sweep_summary.csv
@@ -0,0 +1 @@
+concurrency,qps,ttft_p50_ms,ttft_p95_ms,tpot_p50_ms,system_tps
diff --git a/test_submissions/sub_g/systems/sys_8gpu_vllm.json b/test_submissions/sub_g/sys_8gpu_vllm/system_desc.json
similarity index 100%
rename from test_submissions/sub_g/systems/sys_8gpu_vllm.json
rename to test_submissions/sub_g/sys_8gpu_vllm/system_desc.json
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/accuracy/accuracy.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/accuracy/accuracy_result.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_1024/mlperf_endpoints_log_detail.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_1024/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_1024/mlperf_endpoints_log_summary.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_1024/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_1024.yaml b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/point.yaml
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_1024.yaml
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/point.yaml
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/report.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/report.txt
new file mode 100644
index 0000000..ef6db2b
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/report.txt
@@ -0,0 +1 @@
+Run r1024 — autogenerated report placeholder
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/run_metadata.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/src/vllm/.gitkeep b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1024/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/accuracy/accuracy.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/accuracy/accuracy_result.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_128/mlperf_endpoints_log_detail.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_128/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_128/mlperf_endpoints_log_summary.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_128/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_128.yaml b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/point.yaml
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_128.yaml
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/point.yaml
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/report.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/report.txt
new file mode 100644
index 0000000..2128812
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/report.txt
@@ -0,0 +1 @@
+Run r128 — autogenerated report placeholder
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/run_metadata.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/src/vllm/.gitkeep b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r128/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/accuracy/accuracy.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/accuracy/accuracy_result.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_1536/mlperf_endpoints_log_detail.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_1536/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_1536/mlperf_endpoints_log_summary.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_1536/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_1536.yaml b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/point.yaml
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_1536.yaml
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/point.yaml
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/report.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/report.txt
new file mode 100644
index 0000000..21dc3b9
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/report.txt
@@ -0,0 +1 @@
+Run r1536 — autogenerated report placeholder
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/run_metadata.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/src/vllm/.gitkeep b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r1536/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/accuracy/accuracy.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/accuracy/accuracy_result.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_192/mlperf_endpoints_log_detail.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_192/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_192/mlperf_endpoints_log_summary.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_192/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_192.yaml b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/point.yaml
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_192.yaml
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/point.yaml
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/report.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/report.txt
new file mode 100644
index 0000000..d6d1fdb
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/report.txt
@@ -0,0 +1 @@
+Run r192 — autogenerated report placeholder
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/run_metadata.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/src/vllm/.gitkeep b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r192/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/accuracy/accuracy.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/accuracy/accuracy_result.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_2048/mlperf_endpoints_log_detail.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_2048/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_2048/mlperf_endpoints_log_summary.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_2048/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_2048.yaml b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/point.yaml
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_2048.yaml
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/point.yaml
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/report.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/report.txt
new file mode 100644
index 0000000..d8874f4
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/report.txt
@@ -0,0 +1 @@
+Run r2048 — autogenerated report placeholder
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/run_metadata.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/src/vllm/.gitkeep b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r2048/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/accuracy/accuracy.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/accuracy/accuracy_result.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_256/mlperf_endpoints_log_detail.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_256/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_256/mlperf_endpoints_log_summary.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_256/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_256.yaml b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/point.yaml
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_256.yaml
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/point.yaml
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/report.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/report.txt
new file mode 100644
index 0000000..91b6b15
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/report.txt
@@ -0,0 +1 @@
+Run r256 — autogenerated report placeholder
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/run_metadata.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/src/vllm/.gitkeep b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r256/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/accuracy/accuracy.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/accuracy/accuracy_result.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_384/mlperf_endpoints_log_detail.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_384/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_384/mlperf_endpoints_log_summary.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_384/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_384.yaml b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/point.yaml
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_384.yaml
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/point.yaml
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/report.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/report.txt
new file mode 100644
index 0000000..95d6b85
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/report.txt
@@ -0,0 +1 @@
+Run r384 — autogenerated report placeholder
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/run_metadata.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/src/vllm/.gitkeep b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r384/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/accuracy/accuracy.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/accuracy/accuracy_result.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_512/mlperf_endpoints_log_detail.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_512/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_512/mlperf_endpoints_log_summary.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_512/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_512.yaml b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/point.yaml
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_512.yaml
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/point.yaml
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/report.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/report.txt
new file mode 100644
index 0000000..ea5355f
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/report.txt
@@ -0,0 +1 @@
+Run r512 — autogenerated report placeholder
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/run_metadata.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/src/vllm/.gitkeep b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r512/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/accuracy/accuracy.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/accuracy/accuracy_result.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_64/mlperf_endpoints_log_detail.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_64/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_64/mlperf_endpoints_log_summary.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_64/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_64.yaml b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/point.yaml
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_64.yaml
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/point.yaml
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/report.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/report.txt
new file mode 100644
index 0000000..76f16dd
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/report.txt
@@ -0,0 +1 @@
+Run r64 — autogenerated report placeholder
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/run_metadata.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/src/vllm/.gitkeep b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r64/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/accuracy/accuracy.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/accuracy/accuracy_result.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_768/mlperf_endpoints_log_detail.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_768/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_768/mlperf_endpoints_log_summary.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/results/point_768/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_768.yaml b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/point.yaml
similarity index 100%
rename from test_submissions/sub_h/pareto/sys_8gpu_sglang/llama3-70b/points/point_768.yaml
rename to test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/point.yaml
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/report.txt b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/report.txt
new file mode 100644
index 0000000..80814c9
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/report.txt
@@ -0,0 +1 @@
+Run r768 — autogenerated report placeholder
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/run_metadata.json b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/src/vllm/.gitkeep b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/r768/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/sweep_distributions.csv b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/sweep_distributions.csv
new file mode 100644
index 0000000..f501c1f
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/sweep_distributions.csv
@@ -0,0 +1 @@
+concurrency,percentile,ttft_ms,tpot_ms,output_tokens
diff --git a/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/sweep_summary.csv b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/sweep_summary.csv
new file mode 100644
index 0000000..3faf209
--- /dev/null
+++ b/test_submissions/sub_h/sys_8gpu_sglang/llama3-70b/sweep_summary.csv
@@ -0,0 +1 @@
+concurrency,qps,ttft_p50_ms,ttft_p95_ms,tpot_p50_ms,system_tps
diff --git a/test_submissions/sub_h/systems/sys_8gpu_sglang.json b/test_submissions/sub_h/sys_8gpu_sglang/system_desc.json
similarity index 100%
rename from test_submissions/sub_h/systems/sys_8gpu_sglang.json
rename to test_submissions/sub_h/sys_8gpu_sglang/system_desc.json
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/accuracy/accuracy.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/accuracy/accuracy_result.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_1/mlperf_endpoints_log_detail.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_1/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_1/mlperf_endpoints_log_summary.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_1/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_1.yaml b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/point.yaml
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_1.yaml
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/point.yaml
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/report.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/report.txt
new file mode 100644
index 0000000..bf18a4f
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/report.txt
@@ -0,0 +1 @@
+Run r1 — autogenerated report placeholder
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/run_metadata.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/src/vllm/.gitkeep b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r1/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/accuracy/accuracy.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/accuracy/accuracy_result.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_128/mlperf_endpoints_log_detail.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_128/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_128/mlperf_endpoints_log_summary.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_128/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_128.yaml b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/point.yaml
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_128.yaml
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/point.yaml
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/report.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/report.txt
new file mode 100644
index 0000000..2128812
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/report.txt
@@ -0,0 +1 @@
+Run r128 — autogenerated report placeholder
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/run_metadata.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/src/vllm/.gitkeep b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r128/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/accuracy/accuracy.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/accuracy/accuracy_result.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_16/mlperf_endpoints_log_detail.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_16/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_16/mlperf_endpoints_log_summary.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_16/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_16.yaml b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/point.yaml
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_16.yaml
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/point.yaml
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/report.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/report.txt
new file mode 100644
index 0000000..7268543
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/report.txt
@@ -0,0 +1 @@
+Run r16 — autogenerated report placeholder
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/run_metadata.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/src/vllm/.gitkeep b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r16/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/accuracy/accuracy.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/accuracy/accuracy_result.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_2/mlperf_endpoints_log_detail.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_2/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_2/mlperf_endpoints_log_summary.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_2/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_2.yaml b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/point.yaml
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_2.yaml
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/point.yaml
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/report.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/report.txt
new file mode 100644
index 0000000..a07a826
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/report.txt
@@ -0,0 +1 @@
+Run r2 — autogenerated report placeholder
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/run_metadata.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/src/vllm/.gitkeep b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r2/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/accuracy/accuracy.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/accuracy/accuracy_result.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_256/mlperf_endpoints_log_detail.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_256/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_256/mlperf_endpoints_log_summary.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_256/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_256.yaml b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/point.yaml
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_256.yaml
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/point.yaml
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/report.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/report.txt
new file mode 100644
index 0000000..91b6b15
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/report.txt
@@ -0,0 +1 @@
+Run r256 — autogenerated report placeholder
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/run_metadata.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/src/vllm/.gitkeep b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r256/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/accuracy/accuracy.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/accuracy/accuracy_result.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_32/mlperf_endpoints_log_detail.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_32/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_32/mlperf_endpoints_log_summary.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_32/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_32.yaml b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/point.yaml
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_32.yaml
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/point.yaml
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/report.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/report.txt
new file mode 100644
index 0000000..51064e6
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/report.txt
@@ -0,0 +1 @@
+Run r32 — autogenerated report placeholder
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/run_metadata.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/src/vllm/.gitkeep b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r32/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/accuracy/accuracy.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/accuracy/accuracy_result.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_4/mlperf_endpoints_log_detail.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_4/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_4/mlperf_endpoints_log_summary.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_4/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_4.yaml b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/point.yaml
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_4.yaml
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/point.yaml
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/report.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/report.txt
new file mode 100644
index 0000000..024a6e2
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/report.txt
@@ -0,0 +1 @@
+Run r4 — autogenerated report placeholder
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/run_metadata.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/src/vllm/.gitkeep b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r4/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/accuracy/accuracy.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/accuracy/accuracy_result.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_512/mlperf_endpoints_log_detail.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_512/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_512/mlperf_endpoints_log_summary.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_512/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_512.yaml b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/point.yaml
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_512.yaml
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/point.yaml
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/report.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/report.txt
new file mode 100644
index 0000000..ea5355f
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/report.txt
@@ -0,0 +1 @@
+Run r512 — autogenerated report placeholder
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/run_metadata.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/src/vllm/.gitkeep b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r512/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/accuracy/accuracy.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/accuracy/accuracy_result.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_64/mlperf_endpoints_log_detail.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_64/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_64/mlperf_endpoints_log_summary.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_64/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_64.yaml b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/point.yaml
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_64.yaml
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/point.yaml
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/report.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/report.txt
new file mode 100644
index 0000000..76f16dd
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/report.txt
@@ -0,0 +1 @@
+Run r64 — autogenerated report placeholder
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/run_metadata.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/src/vllm/.gitkeep b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r64/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/accuracy/accuracy.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/accuracy/accuracy_result.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_8/mlperf_endpoints_log_detail.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_8/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_8/mlperf_endpoints_log_summary.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/results/point_8/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_8.yaml b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/point.yaml
similarity index 100%
rename from test_submissions/sub_i/pareto/sys_h200_8gpu/deepseek-r1/points/point_8.yaml
rename to test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/point.yaml
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/report.txt b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/report.txt
new file mode 100644
index 0000000..e4c38ca
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/report.txt
@@ -0,0 +1 @@
+Run r8 — autogenerated report placeholder
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/run_metadata.json b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/src/vllm/.gitkeep b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/r8/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/sweep_distributions.csv b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/sweep_distributions.csv
new file mode 100644
index 0000000..f501c1f
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/sweep_distributions.csv
@@ -0,0 +1 @@
+concurrency,percentile,ttft_ms,tpot_ms,output_tokens
diff --git a/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/sweep_summary.csv b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/sweep_summary.csv
new file mode 100644
index 0000000..3faf209
--- /dev/null
+++ b/test_submissions/sub_i/sys_h200_8gpu/deepseek-r1/sweep_summary.csv
@@ -0,0 +1 @@
+concurrency,qps,ttft_p50_ms,ttft_p95_ms,tpot_p50_ms,system_tps
diff --git a/test_submissions/sub_i/systems/sys_h200_8gpu.json b/test_submissions/sub_i/sys_h200_8gpu/system_desc.json
similarity index 100%
rename from test_submissions/sub_i/systems/sys_h200_8gpu.json
rename to test_submissions/sub_i/sys_h200_8gpu/system_desc.json
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/accuracy/accuracy.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/accuracy/accuracy_result.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_1024/mlperf_endpoints_log_detail.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_1024/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_1024/mlperf_endpoints_log_summary.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_1024/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_1024.yaml b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/point.yaml
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_1024.yaml
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/point.yaml
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/report.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/report.txt
new file mode 100644
index 0000000..ef6db2b
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/report.txt
@@ -0,0 +1 @@
+Run r1024 — autogenerated report placeholder
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/run_metadata.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/src/vllm/.gitkeep b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r1024/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/accuracy/accuracy.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/accuracy/accuracy_result.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_128/mlperf_endpoints_log_detail.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_128/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_128/mlperf_endpoints_log_summary.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_128/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_128.yaml b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/point.yaml
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_128.yaml
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/point.yaml
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/report.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/report.txt
new file mode 100644
index 0000000..2128812
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/report.txt
@@ -0,0 +1 @@
+Run r128 — autogenerated report placeholder
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/run_metadata.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/src/vllm/.gitkeep b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r128/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/accuracy/accuracy.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/accuracy/accuracy_result.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_16384/mlperf_endpoints_log_detail.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_16384/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_16384/mlperf_endpoints_log_summary.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_16384/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_16384.yaml b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/point.yaml
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_16384.yaml
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/point.yaml
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/report.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/report.txt
new file mode 100644
index 0000000..0752e25
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/report.txt
@@ -0,0 +1 @@
+Run r16384 — autogenerated report placeholder
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/run_metadata.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/src/vllm/.gitkeep b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r16384/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/accuracy/accuracy.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/accuracy/accuracy_result.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_2048/mlperf_endpoints_log_detail.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_2048/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_2048/mlperf_endpoints_log_summary.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_2048/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_2048.yaml b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/point.yaml
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_2048.yaml
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/point.yaml
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/report.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/report.txt
new file mode 100644
index 0000000..d8874f4
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/report.txt
@@ -0,0 +1 @@
+Run r2048 — autogenerated report placeholder
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/run_metadata.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/src/vllm/.gitkeep b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r2048/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/accuracy/accuracy.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/accuracy/accuracy_result.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_256/mlperf_endpoints_log_detail.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_256/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_256/mlperf_endpoints_log_summary.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_256/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_256.yaml b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/point.yaml
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_256.yaml
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/point.yaml
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/report.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/report.txt
new file mode 100644
index 0000000..91b6b15
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/report.txt
@@ -0,0 +1 @@
+Run r256 — autogenerated report placeholder
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/run_metadata.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/src/vllm/.gitkeep b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r256/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/accuracy/accuracy.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/accuracy/accuracy_result.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_32/mlperf_endpoints_log_detail.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_32/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_32/mlperf_endpoints_log_summary.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_32/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_32.yaml b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/point.yaml
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_32.yaml
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/point.yaml
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/report.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/report.txt
new file mode 100644
index 0000000..51064e6
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/report.txt
@@ -0,0 +1 @@
+Run r32 — autogenerated report placeholder
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/run_metadata.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/src/vllm/.gitkeep b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r32/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/accuracy/accuracy.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/accuracy/accuracy_result.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_4096/mlperf_endpoints_log_detail.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_4096/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_4096/mlperf_endpoints_log_summary.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_4096/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_4096.yaml b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/point.yaml
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_4096.yaml
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/point.yaml
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/report.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/report.txt
new file mode 100644
index 0000000..c617ae7
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/report.txt
@@ -0,0 +1 @@
+Run r4096 — autogenerated report placeholder
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/run_metadata.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/src/vllm/.gitkeep b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r4096/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/accuracy/accuracy.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/accuracy/accuracy_result.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_512/mlperf_endpoints_log_detail.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_512/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_512/mlperf_endpoints_log_summary.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_512/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_512.yaml b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/point.yaml
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_512.yaml
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/point.yaml
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/report.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/report.txt
new file mode 100644
index 0000000..ea5355f
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/report.txt
@@ -0,0 +1 @@
+Run r512 — autogenerated report placeholder
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/run_metadata.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/src/vllm/.gitkeep b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r512/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/accuracy/accuracy.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/accuracy/accuracy_result.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_64/mlperf_endpoints_log_detail.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_64/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_64/mlperf_endpoints_log_summary.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_64/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_64.yaml b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/point.yaml
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_64.yaml
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/point.yaml
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/report.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/report.txt
new file mode 100644
index 0000000..76f16dd
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/report.txt
@@ -0,0 +1 @@
+Run r64 — autogenerated report placeholder
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/run_metadata.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/src/vllm/.gitkeep b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r64/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/accuracy/accuracy.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/accuracy/accuracy.txt
new file mode 100644
index 0000000..9a87ee9
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/accuracy/accuracy.txt
@@ -0,0 +1 @@
+rouge1: 0.45
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/accuracy/accuracy_result.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/accuracy/accuracy_result.json
new file mode 100644
index 0000000..1a39c7f
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.45,
+  "quality_target": 0.43,
+  "passed": true
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_8192/mlperf_endpoints_log_detail.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_8192/mlperf_endpoints_log_detail.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_8192/mlperf_endpoints_log_summary.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/results/point_8192/mlperf_endpoints_log_summary.json
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_8192.yaml b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/point.yaml
similarity index 100%
rename from test_submissions/sub_j/pareto/sys_gb300_72gpu/deepseek-r1/points/point_8192.yaml
rename to test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/point.yaml
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/report.txt b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/report.txt
new file mode 100644
index 0000000..2423880
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/report.txt
@@ -0,0 +1 @@
+Run r8192 — autogenerated report placeholder
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/run_metadata.json b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/src/vllm/.gitkeep b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/r8192/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/sweep_distributions.csv b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/sweep_distributions.csv
new file mode 100644
index 0000000..f501c1f
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/sweep_distributions.csv
@@ -0,0 +1 @@
+concurrency,percentile,ttft_ms,tpot_ms,output_tokens
diff --git a/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/sweep_summary.csv b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/sweep_summary.csv
new file mode 100644
index 0000000..3faf209
--- /dev/null
+++ b/test_submissions/sub_j/sys_gb300_72gpu/deepseek-r1/sweep_summary.csv
@@ -0,0 +1 @@
+concurrency,qps,ttft_p50_ms,ttft_p95_ms,tpot_p50_ms,system_tps
diff --git a/test_submissions/sub_j/systems/sys_gb300_72gpu.json b/test_submissions/sub_j/sys_gb300_72gpu/system_desc.json
similarity index 100%
rename from test_submissions/sub_j/systems/sys_gb300_72gpu.json
rename to test_submissions/sub_j/sys_gb300_72gpu/system_desc.json
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/accuracy/accuracy.txt b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/accuracy/accuracy.txt
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/accuracy/accuracy.txt
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/accuracy/accuracy.txt
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/accuracy/accuracy_result.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/accuracy/accuracy_result.json
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/accuracy/accuracy_result.json
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/accuracy/accuracy_result.json
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_1000/mlperf_endpoints_log_detail.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_1000/mlperf_endpoints_log_detail.json
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_1000/mlperf_endpoints_log_summary.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_1000/mlperf_endpoints_log_summary.json
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/points/point_1000.yaml b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/point.yaml
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/points/point_1000.yaml
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/point.yaml
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/report.txt b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/report.txt
new file mode 100644
index 0000000..8981711
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/report.txt
@@ -0,0 +1 @@
+Run r1000 — autogenerated report placeholder
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/run_metadata.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/src/vllm/.gitkeep b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r1000/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/accuracy/accuracy.txt b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/accuracy/accuracy.txt
new file mode 100644
index 0000000..ad5d096
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/accuracy/accuracy.txt
@@ -0,0 +1,3 @@
+rouge1: 0.4512
+rouge2: 0.2201
+rougeL: 0.3817
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/accuracy/accuracy_result.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/accuracy/accuracy_result.json
new file mode 100644
index 0000000..7ebab9b
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.4512,
+  "quality_target": 0.43,
+  "passed": true
+}
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_16/mlperf_endpoints_log_detail.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_16/mlperf_endpoints_log_detail.json
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_16/mlperf_endpoints_log_summary.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_16/mlperf_endpoints_log_summary.json
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/points/point_16.yaml b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/point.yaml
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/points/point_16.yaml
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/point.yaml
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/report.txt b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/report.txt
new file mode 100644
index 0000000..7268543
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/report.txt
@@ -0,0 +1 @@
+Run r16 — autogenerated report placeholder
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/run_metadata.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/src/vllm/.gitkeep b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r16/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/accuracy/accuracy.txt b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/accuracy/accuracy.txt
new file mode 100644
index 0000000..ad5d096
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/accuracy/accuracy.txt
@@ -0,0 +1,3 @@
+rouge1: 0.4512
+rouge2: 0.2201
+rougeL: 0.3817
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/accuracy/accuracy_result.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/accuracy/accuracy_result.json
new file mode 100644
index 0000000..7ebab9b
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.4512,
+  "quality_target": 0.43,
+  "passed": true
+}
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_256/mlperf_endpoints_log_detail.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_256/mlperf_endpoints_log_detail.json
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_256/mlperf_endpoints_log_summary.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_256/mlperf_endpoints_log_summary.json
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/points/point_256.yaml b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/point.yaml
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/points/point_256.yaml
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/point.yaml
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/report.txt b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/report.txt
new file mode 100644
index 0000000..91b6b15
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/report.txt
@@ -0,0 +1 @@
+Run r256 — autogenerated report placeholder
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/run_metadata.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/src/vllm/.gitkeep b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r256/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/accuracy/accuracy.txt b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/accuracy/accuracy.txt
new file mode 100644
index 0000000..ad5d096
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/accuracy/accuracy.txt
@@ -0,0 +1,3 @@
+rouge1: 0.4512
+rouge2: 0.2201
+rougeL: 0.3817
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/accuracy/accuracy_result.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/accuracy/accuracy_result.json
new file mode 100644
index 0000000..7ebab9b
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.4512,
+  "quality_target": 0.43,
+  "passed": true
+}
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_38/mlperf_endpoints_log_detail.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_38/mlperf_endpoints_log_detail.json
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_38/mlperf_endpoints_log_summary.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_38/mlperf_endpoints_log_summary.json
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/points/point_38.yaml b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/point.yaml
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/points/point_38.yaml
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/point.yaml
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/report.txt b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/report.txt
new file mode 100644
index 0000000..e151c43
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/report.txt
@@ -0,0 +1 @@
+Run r38 — autogenerated report placeholder
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/run_metadata.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/src/vllm/.gitkeep b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r38/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/accuracy/accuracy.txt b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/accuracy/accuracy.txt
new file mode 100644
index 0000000..ad5d096
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/accuracy/accuracy.txt
@@ -0,0 +1,3 @@
+rouge1: 0.4512
+rouge2: 0.2201
+rougeL: 0.3817
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/accuracy/accuracy_result.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/accuracy/accuracy_result.json
new file mode 100644
index 0000000..7ebab9b
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.4512,
+  "quality_target": 0.43,
+  "passed": true
+}
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_512/mlperf_endpoints_log_detail.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_512/mlperf_endpoints_log_detail.json
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_512/mlperf_endpoints_log_summary.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_512/mlperf_endpoints_log_summary.json
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/points/point_512.yaml b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/point.yaml
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/points/point_512.yaml
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/point.yaml
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/report.txt b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/report.txt
new file mode 100644
index 0000000..ea5355f
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/report.txt
@@ -0,0 +1 @@
+Run r512 — autogenerated report placeholder
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/run_metadata.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/src/vllm/.gitkeep b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r512/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/accuracy/accuracy.txt b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/accuracy/accuracy.txt
new file mode 100644
index 0000000..ad5d096
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/accuracy/accuracy.txt
@@ -0,0 +1,3 @@
+rouge1: 0.4512
+rouge2: 0.2201
+rougeL: 0.3817
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/accuracy/accuracy_result.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/accuracy/accuracy_result.json
new file mode 100644
index 0000000..7ebab9b
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.4512,
+  "quality_target": 0.43,
+  "passed": true
+}
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_768/mlperf_endpoints_log_detail.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_768/mlperf_endpoints_log_detail.json
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_768/mlperf_endpoints_log_summary.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_768/mlperf_endpoints_log_summary.json
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/points/point_768.yaml b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/point.yaml
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/points/point_768.yaml
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/point.yaml
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/report.txt b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/report.txt
new file mode 100644
index 0000000..80814c9
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/report.txt
@@ -0,0 +1 @@
+Run r768 — autogenerated report placeholder
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/run_metadata.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/src/vllm/.gitkeep b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r768/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/accuracy/accuracy.txt b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/accuracy/accuracy.txt
new file mode 100644
index 0000000..ad5d096
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/accuracy/accuracy.txt
@@ -0,0 +1,3 @@
+rouge1: 0.4512
+rouge2: 0.2201
+rougeL: 0.3817
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/accuracy/accuracy_result.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/accuracy/accuracy_result.json
new file mode 100644
index 0000000..7ebab9b
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/accuracy/accuracy_result.json
@@ -0,0 +1,6 @@
+{
+  "metric": "rouge1",
+  "score": 0.4512,
+  "quality_target": 0.43,
+  "passed": true
+}
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_88/mlperf_endpoints_log_detail.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/mlperf_endpoints_log_detail.json
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_88/mlperf_endpoints_log_detail.json
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/mlperf_endpoints_log_detail.json
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_88/mlperf_endpoints_log_summary.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/mlperf_endpoints_log_summary.json
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/results/point_88/mlperf_endpoints_log_summary.json
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/mlperf_endpoints_log_summary.json
diff --git a/test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/points/point_88.yaml b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/point.yaml
similarity index 100%
rename from test_submissions/valid_standardized/pareto/acme_h100x8_001/llama3-70b/points/point_88.yaml
rename to test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/point.yaml
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/report.txt b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/report.txt
new file mode 100644
index 0000000..cbb894a
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/report.txt
@@ -0,0 +1 @@
+Run r88 — autogenerated report placeholder
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/run_metadata.json b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/run_metadata.json
new file mode 100644
index 0000000..39442b5
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/run_metadata.json
@@ -0,0 +1,5 @@
+{
+  "serving_framework": "vLLM",
+  "parallelism": {},
+  "precision": "bfloat16"
+}
\ No newline at end of file
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/src/vllm/.gitkeep b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/r88/src/vllm/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/sweep_distributions.csv b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/sweep_distributions.csv
new file mode 100644
index 0000000..f501c1f
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/sweep_distributions.csv
@@ -0,0 +1 @@
+concurrency,percentile,ttft_ms,tpot_ms,output_tokens
diff --git a/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/sweep_summary.csv b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/sweep_summary.csv
new file mode 100644
index 0000000..3faf209
--- /dev/null
+++ b/test_submissions/valid_standardized/acme_h100x8_001/llama3-70b/sweep_summary.csv
@@ -0,0 +1 @@
+concurrency,qps,ttft_p50_ms,ttft_p95_ms,tpot_p50_ms,system_tps
diff --git a/test_submissions/valid_standardized/systems/acme_h100x8_001.json b/test_submissions/valid_standardized/acme_h100x8_001/system_desc.json
similarity index 100%
rename from test_submissions/valid_standardized/systems/acme_h100x8_001.json
rename to test_submissions/valid_standardized/acme_h100x8_001/system_desc.json
diff --git a/tests/endpoints_submission_cli/submissions/test_builder.py b/tests/endpoints_submission_cli/submissions/test_builder.py
index 01fde01..ab8eb2b 100644
--- a/tests/endpoints_submission_cli/submissions/test_builder.py
+++ b/tests/endpoints_submission_cli/submissions/test_builder.py
@@ -43,26 +43,29 @@ def test_creates_dest_dir(self, run_archive: Path, tmp_path: Path) -> None:
 
 @pytest.mark.unit
 class TestBuildSubmissionFolder:
-    def test_creates_systems_dir(self, run_archive: Path, tmp_path: Path) -> None:
+    def test_creates_system_desc_json(self, run_archive: Path, tmp_path: Path) -> None:
         sub_dir = build_submission_folder(
             [("run-001", run_archive)], "standardized", tmp_path
         )
-        assert (sub_dir / "systems").is_dir()
+        system_descs = list(sub_dir.rglob("system_desc.json"))
+        assert len(system_descs) == 1
 
-    def test_creates_pareto_dir(self, run_archive: Path, tmp_path: Path) -> None:
+    def test_creates_model_dir(self, run_archive: Path, tmp_path: Path) -> None:
         sub_dir = build_submission_folder(
             [("run-001", run_archive)], "standardized", tmp_path
         )
-        pareto = sub_dir / "pareto"
-        assert pareto.is_dir()
+        # <system>/<model>/ should exist
+        system_dirs = [d for d in sub_dir.iterdir() if d.is_dir()]
+        assert len(system_dirs) >= 1
+        model_dirs = [d for d in system_dirs[0].iterdir() if d.is_dir()]
+        assert len(model_dirs) >= 1
 
-    def test_system_json_created(self, run_archive: Path, tmp_path: Path) -> None:
+    def test_system_desc_json_content(self, run_archive: Path, tmp_path: Path) -> None:
         sub_dir = build_submission_folder(
             [("run-001", run_archive)], "standardized", tmp_path
         )
-        jsons = list((sub_dir / "systems").glob("*.json"))
-        assert len(jsons) == 1
-        data = json.loads(jsons[0].read_text())
+        descs = list(sub_dir.rglob("system_desc.json"))
+        data = json.loads(descs[0].read_text())
         assert "division" in data
         assert data["division"] == "Standardized"
 
@@ -70,7 +73,7 @@ def test_point_yaml_created(self, run_archive: Path, tmp_path: Path) -> None:
         sub_dir = build_submission_folder(
             [("run-001", run_archive)], "standardized", tmp_path
         )
-        yamls = list(sub_dir.rglob("point_*.yaml"))
+        yamls = list(sub_dir.rglob("point.yaml"))
         assert len(yamls) >= 1
         data = yaml.safe_load(yamls[0].read_text())
         assert data["concurrency"] == 4
@@ -92,6 +95,36 @@ def test_log_detail_created(self, run_archive: Path, tmp_path: Path) -> None:
         details = list(sub_dir.rglob("mlperf_endpoints_log_detail.json"))
         assert len(details) == 1
 
+    def test_run_metadata_created(self, run_archive: Path, tmp_path: Path) -> None:
+        sub_dir = build_submission_folder(
+            [("run-001", run_archive)], "standardized", tmp_path
+        )
+        assert len(list(sub_dir.rglob("run_metadata.json"))) == 1
+
+    def test_report_txt_created(self, run_archive: Path, tmp_path: Path) -> None:
+        sub_dir = build_submission_folder(
+            [("run-001", run_archive)], "standardized", tmp_path
+        )
+        assert len(list(sub_dir.rglob("report.txt"))) == 1
+
+    def test_sweep_csvs_created(self, run_archive: Path, tmp_path: Path) -> None:
+        sub_dir = build_submission_folder(
+            [("run-001", run_archive)], "standardized", tmp_path
+        )
+        assert len(list(sub_dir.rglob("sweep_summary.csv"))) == 1
+        assert len(list(sub_dir.rglob("sweep_distributions.csv"))) == 1
+
+    def test_src_impl_dir_created(self, run_archive: Path, tmp_path: Path) -> None:
+        sub_dir = build_submission_folder(
+            [("run-001", run_archive)], "standardized", tmp_path
+        )
+        src_dirs = [d for d in sub_dir.rglob("src") if d.is_dir()]
+        assert len(src_dirs) >= 1
+        # Each src/ must have at least one implementation subdir
+        for src in src_dirs:
+            impl_dirs = [d for d in src.iterdir() if d.is_dir()]
+            assert len(impl_dirs) >= 1
+
     def test_accuracy_files_created(self, run_archive: Path, tmp_path: Path) -> None:
         sub_dir = build_submission_folder(
             [("run-001", run_archive)], "standardized", tmp_path
@@ -101,6 +134,15 @@ def test_accuracy_files_created(self, run_archive: Path, tmp_path: Path) -> None
         assert len(acc_txts) == 1
         assert len(acc_jsons) == 1
 
+    def test_run_dir_pattern(self, run_archive: Path, tmp_path: Path) -> None:
+        """Run directories follow the r<N> naming convention."""
+        sub_dir = build_submission_folder(
+            [("run-001", run_archive)], "standardized", tmp_path
+        )
+        import re
+        run_dirs = [d for d in sub_dir.rglob("*") if d.is_dir() and re.match(r"^r\d+$", d.name)]
+        assert len(run_dirs) >= 1
+
     def test_empty_run_list_raises(self, tmp_path: Path) -> None:
         with pytest.raises(SubmissionBuildError, match="At least one"):
             build_submission_folder([], "standardized", tmp_path)
@@ -143,17 +185,21 @@ def test_multiple_runs_single_system(
             "standardized",
             tmp_path / "sub",
         )
-        yamls = list(sub_dir.rglob("point_*.yaml"))
-        concurrencies = {yaml.safe_load(p.read_text())["concurrency"] for p in yamls}
-        assert 4 in concurrencies
-        assert 16 in concurrencies
+        import re
+        run_dirs = {
+            d.name
+            for d in sub_dir.rglob("*")
+            if d.is_dir() and re.match(r"^r\d+$", d.name)
+        }
+        assert "r4" in run_dirs
+        assert "r16" in run_dirs
 
     def test_serviced_division_normalized(self, run_archive: Path, tmp_path: Path) -> None:
         sub_dir = build_submission_folder(
             [("run-001", run_archive)], "serviced", tmp_path
         )
-        jsons = list((sub_dir / "systems").glob("*.json"))
-        data = json.loads(jsons[0].read_text())
+        descs = list(sub_dir.rglob("system_desc.json"))
+        data = json.loads(descs[0].read_text())
         assert data["division"] == "Serviced"
 
 
@@ -167,7 +213,7 @@ def test_creates_archive(self, run_archive: Path, tmp_path: Path) -> None:
         assert bundle.exists()
         with tarfile.open(bundle) as tar:
             names = tar.getnames()
-        assert any("systems" in n for n in names)
+        assert any("system_desc.json" in n for n in names)
 
     def test_default_dest(self, run_archive: Path, tmp_path: Path) -> None:
         sub_dir = build_submission_folder(
diff --git a/tests/endpoints_submission_cli/submissions/test_github.py b/tests/endpoints_submission_cli/submissions/test_github.py
index ca77142..3277beb 100644
--- a/tests/endpoints_submission_cli/submissions/test_github.py
+++ b/tests/endpoints_submission_cli/submissions/test_github.py
@@ -169,7 +169,10 @@ def test_returns_repo_dir_and_org_dir(self, tmp_path: Path) -> None:
 
     def test_fresh_org_dir_copies_entire_tree(self, tmp_path: Path) -> None:
         submission_dir = tmp_path / "ORG"
-        (submission_dir / "systems").mkdir(parents=True)
+        # New layout: system dir with system_desc.json directly at org root
+        sys_dir = submission_dir / "test-sys"
+        sys_dir.mkdir(parents=True)
+        (sys_dir / "system_desc.json").write_text("{}")
         work_dir = tmp_path / "work"
         (work_dir / "repo").mkdir(parents=True)
 
diff --git a/tests/submission_checker/conftest.py b/tests/submission_checker/conftest.py
index f5437c9..69b1050 100644
--- a/tests/submission_checker/conftest.py
+++ b/tests/submission_checker/conftest.py
@@ -29,6 +29,7 @@
     SystemDescription,
     compute_regions,
 )
+import json
 
 # ---------------------------------------------------------------------------
 # Shared constants
@@ -119,24 +120,19 @@ def _model_ctx(
     loaded_points: list[tuple[PointConfig, PointSummary]] | None = None,
     system_desc: SystemDescription | None = None,
     model_name: str = "llama3-70b",
-    accuracy_result: AccuracyResult | None = None,
+    accuracy_results: list[AccuracyResult | None] | None = None,
 ) -> ModelContext:
     model_dir = tmp_path / model_name
     model_dir.mkdir(exist_ok=True)
-    (model_dir / "points").mkdir(exist_ok=True)
-    (model_dir / "results").mkdir(exist_ok=True)
-    (model_dir / "accuracy").mkdir(exist_ok=True)
     return ModelContext(
         system_id="test-sys",
         system_desc=system_desc or _system_desc(benchmark_model=model_name),
         model_dir=model_dir,
         regions=_REGIONS,
-        points_dir=model_dir / "points",
-        accuracy_dir=model_dir / "accuracy",
         all_point_count=all_point_count,
         valid_points=valid_points or [],
         loaded_points=loaded_points or [],
-        accuracy_result=accuracy_result,
+        accuracy_results=accuracy_results or [],
     )
 
 TEST_SUBMISSIONS = Path(__file__).parent.parent.parent / "test_submissions"
diff --git a/tests/submission_checker/test_checker.py b/tests/submission_checker/test_checker.py
index 75ed5e2..a88f14b 100644
--- a/tests/submission_checker/test_checker.py
+++ b/tests/submission_checker/test_checker.py
@@ -323,6 +323,10 @@ def _build_submission(
     write_results: bool = True,
     write_accuracy: bool = True,
     write_accuracy_json: bool = True,
+    write_sweep_files: bool = True,
+    write_run_metadata: bool = True,
+    write_report: bool = True,
+    write_src: bool = True,
     accuracy_data: dict | None = None,
     model: str = "llama3-70b",
 ) -> Path:
@@ -331,34 +335,46 @@ def _build_submission(
     desc["benchmark_model"] = model
     concs = concurrencies if concurrencies is not None else _CONCURRENCIES
 
-    systems_dir = root / "systems"
-    systems_dir.mkdir(parents=True)
-    (systems_dir / f"{system_id}.json").write_text(json.dumps(desc))
+    system_dir = root / system_id
+    system_dir.mkdir(parents=True)
+    (system_dir / "system_desc.json").write_text(json.dumps(desc))
 
-    pareto_dir = root / "pareto"
-    model_dir = pareto_dir / system_id / model
-    points_dir = model_dir / "points"
-    results_dir = model_dir / "results"
-    accuracy_dir = model_dir / "accuracy"
-    for d in (points_dir, results_dir, accuracy_dir):
-        d.mkdir(parents=True)
+    model_dir = system_dir / model
+    model_dir.mkdir(parents=True)
+
+    if write_sweep_files:
+        (model_dir / "sweep_summary.csv").write_text("concurrency,qps\n")
+        (model_dir / "sweep_distributions.csv").write_text("concurrency,percentile\n")
 
     if write_runs:
         for c in concs:
-            (points_dir / f"point_{c}.yaml").write_text(yaml.dump(_make_run_yaml(c)))
+            run_dir = model_dir / f"r{c}"
+            run_dir.mkdir(parents=True)
 
-    if write_results:
-        for c in concs:
-            result_dir = results_dir / f"point_{c}"
-            result_dir.mkdir(parents=True)
-            (result_dir / "mlperf_endpoints_log_summary.json").write_text(json.dumps(_SUMMARY))
-            (result_dir / "mlperf_endpoints_log_detail.json").write_text("{}")
+            (run_dir / "point.yaml").write_text(yaml.dump(_make_run_yaml(c)))
+
+            if write_results:
+                (run_dir / "mlperf_endpoints_log_summary.json").write_text(json.dumps(_SUMMARY))
+                (run_dir / "mlperf_endpoints_log_detail.json").write_text("{}")
+
+            if write_run_metadata:
+                (run_dir / "run_metadata.json").write_text("{}")
 
-    if write_accuracy:
-        (accuracy_dir / "accuracy.txt").write_text("ROUGE-1: 0.45")
-    if write_accuracy_json:
-        data = accuracy_data if accuracy_data is not None else _ACCURACY
-        (accuracy_dir / "accuracy_result.json").write_text(json.dumps(data))
+            if write_report:
+                (run_dir / "report.txt").write_text(f"Run r{c}\n")
+
+            if write_src:
+                src_dir = run_dir / "src" / "vllm"
+                src_dir.mkdir(parents=True)
+                (src_dir / ".gitkeep").write_text("")
+
+            acc_dir = run_dir / "accuracy"
+            acc_dir.mkdir()
+            if write_accuracy:
+                (acc_dir / "accuracy.txt").write_text("ROUGE-1: 0.45")
+            if write_accuracy_json:
+                data = accuracy_data if accuracy_data is not None else _ACCURACY
+                (acc_dir / "accuracy_result.json").write_text(json.dumps(data))
 
     return root
 
@@ -371,67 +387,49 @@ def test_nonexistent_path(self, tmp_path):
         report = _check(tmp_path / "does_not_exist")
         assert _errors(report, "path-exists")
 
-    def test_missing_required_dirs_early_exit(self, tmp_path):
-        """SubmissionDir structure errors cause early return from run()."""
-        # Only systems/ present — pareto/ missing → structure error → early exit
-        (tmp_path / "systems").mkdir()
+    def test_no_system_dirs_early_exit(self, tmp_path):
+        """system-dir-present error when no directories contain system_desc.json."""
+        # Only docs/ present — no system_desc.json anywhere
+        (tmp_path / "docs").mkdir()
         report = _check(tmp_path)
-        assert _errors(report, "required-dir")
+        assert _errors(report, "system-dir-present")
         # Should not have processed any systems
-        assert not any(r.rule == "system-description-present" for r in report.results)
-
-    def test_no_system_json_files(self, tmp_path):
-        """system-description-present error when systems/ has no *.json files."""
-        (tmp_path / "systems").mkdir()
-        (tmp_path / "pareto").mkdir()
-        report = _check(tmp_path)
-        assert _errors(report, "system-description-present")
+        assert not any(r.rule == "system-description-valid" for r in report.results)
 
     def test_invalid_system_json(self, tmp_path):
-        """system-description-valid error when system JSON is malformed."""
-        (tmp_path / "systems").mkdir()
-        (tmp_path / "pareto").mkdir()
-        (tmp_path / "systems" / "bad-sys.json").write_text("{bad json")
+        """system-description-valid error when system_desc.json is malformed."""
+        sys_dir = tmp_path / "test-sys"
+        sys_dir.mkdir()
+        (sys_dir / "system_desc.json").write_text("{bad json")
         report = _check(tmp_path)
         assert _errors(report, "system-description-valid")
 
-    def test_missing_pareto_system_dir_early_exit(self, tmp_path):
-        """pareto-dir-exists error when pareto/<system_id>/ is absent."""
-        (tmp_path / "systems").mkdir()
-        (tmp_path / "pareto").mkdir()
-        (tmp_path / "systems" / "test-sys.json").write_text(json.dumps(_SYSTEM_DESC))
-        report = _check(tmp_path)
-        assert _errors(report, "pareto-dir-exists")
-
-    def test_empty_pareto_system_dir(self, tmp_path):
-        """benchmark-model-dir error when pareto/<system_id>/ has no subdirectories."""
-        (tmp_path / "systems").mkdir()
-        (tmp_path / "systems" / "test-sys.json").write_text(json.dumps(_SYSTEM_DESC))
-        pareto_sys = tmp_path / "pareto" / "test-sys"
-        pareto_sys.mkdir(parents=True)
+    def test_no_model_dirs_early_exit(self, tmp_path):
+        """benchmark-model-dir error when system dir has no model subdirectories."""
+        sys_dir = tmp_path / "test-sys"
+        sys_dir.mkdir()
+        (sys_dir / "system_desc.json").write_text(json.dumps(_SYSTEM_DESC))
+        # No model subdirs
         report = _check(tmp_path)
         assert _errors(report, "benchmark-model-dir")
 
-    def test_missing_model_subdirs_early_exit(self, tmp_path):
-        """pareto-subdir error when points/ or results/ or accuracy/ is absent."""
-        (tmp_path / "systems").mkdir()
-        (tmp_path / "systems" / "test-sys.json").write_text(json.dumps(_SYSTEM_DESC))
-        model_dir = tmp_path / "pareto" / "test-sys" / "llama3-70b"
-        # Only points/ present — results/ and accuracy/ missing
-        (model_dir / "points").mkdir(parents=True)
+    def test_docs_dir_not_treated_as_model(self, tmp_path):
+        """docs/ inside a system dir is not treated as a model directory."""
+        sys_dir = tmp_path / "test-sys"
+        sys_dir.mkdir()
+        (sys_dir / "system_desc.json").write_text(json.dumps(_SYSTEM_DESC))
+        (sys_dir / "docs").mkdir()  # should be ignored as a model dir
         report = _check(tmp_path)
-        assert _errors(report, "pareto-subdir")
-        # Should not attempt to list point_*.yaml (early exit after structure errors)
-        assert not any(r.rule == "measurement-points-present" for r in report.results)
+        assert _errors(report, "benchmark-model-dir")
 
-    def test_no_point_yamls(self, tmp_path):
-        """measurement-points-present error when points/ has no point_*.yaml files."""
-        root = _build_submission(tmp_path, write_runs=False, write_results=False)
+    def test_no_run_dirs(self, tmp_path):
+        """measurement-points-present error when model dir has no r<N>/ directories."""
+        root = _build_submission(tmp_path, write_runs=False)
         report = _check(root)
         assert _errors(report, "measurement-points-present")
 
     def test_missing_result_log(self, tmp_path):
-        """result-file-present error when results/point_<N>/ log is absent."""
+        """result-file-present error when mlperf_endpoints_log_summary.json is absent."""
         root = _build_submission(tmp_path, write_results=False)
         report = _check(root)
         assert _errors(report, "result-file-present")
@@ -439,8 +437,8 @@ def test_missing_result_log(self, tmp_path):
     def test_missing_detail_log(self, tmp_path):
         """result-detail-present error when mlperf_endpoints_log_detail.json is absent."""
         root = _build_submission(tmp_path)
-        # Remove the detail log for one point
-        detail = root / "pareto" / "test-sys" / "llama3-70b" / "results" / "point_16" / "mlperf_endpoints_log_detail.json"
+        # Remove the detail log for one run
+        detail = root / "test-sys" / "llama3-70b" / "r16" / "mlperf_endpoints_log_detail.json"
         detail.unlink()
         report = _check(root)
         assert _errors(report, "result-detail-present")
@@ -448,11 +446,8 @@ def test_missing_detail_log(self, tmp_path):
     def test_invalid_result_log(self, tmp_path):
         """result-file-valid error when the result log JSON is malformed."""
         root = _build_submission(tmp_path)
-        # Overwrite one summary with invalid JSON
-        bad_path = root / "pareto" / "test-sys" / "llama3-70b" / "results" / "point_16"
-        bad_path.mkdir(parents=True, exist_ok=True)
-        (bad_path / "mlperf_endpoints_log_summary.json").write_text("{bad")
-        (bad_path / "mlperf_endpoints_log_detail.json").write_text("{}")
+        bad_path = root / "test-sys" / "llama3-70b" / "r16" / "mlperf_endpoints_log_summary.json"
+        bad_path.write_text("{bad")
         report = _check(root)
         assert _errors(report, "result-file-valid")
 
@@ -477,56 +472,56 @@ def test_invalid_accuracy_json(self, tmp_path):
         report = _check(root)
         assert _errors(report, "accuracy-valid")
 
-    def test_point_filename_concurrency_mismatch(self, tmp_path):
-        """point-filename-concurrency warning when filename concurrency ≠ declared concurrency."""
+    def test_dir_concurrency_mismatch_warning(self, tmp_path):
+        """point-filename-concurrency warning when r<N>/ dir concurrency ≠ declared concurrency."""
         root = _build_submission(tmp_path)
-        # Add a point file whose name says 999 but YAML declares 64
-        mismatch_yaml = root / "pareto" / "test-sys" / "llama3-70b" / "points" / "point_999.yaml"
-        mismatch_yaml.write_text(yaml.dump(_make_run_yaml(64)))
-        # Also add the matching result dir so it doesn't error on result-file-present
-        result_dir = root / "pareto" / "test-sys" / "llama3-70b" / "results" / "point_64"
-        result_dir.mkdir(parents=True, exist_ok=True)
-        (result_dir / "mlperf_endpoints_log_summary.json").write_text(json.dumps(_SUMMARY))
-        (result_dir / "mlperf_endpoints_log_detail.json").write_text("{}")
+        # Add a run dir r999 that declares concurrency 64
+        mismatch_dir = root / "test-sys" / "llama3-70b" / "r999"
+        mismatch_dir.mkdir()
+        (mismatch_dir / "point.yaml").write_text(yaml.dump(_make_run_yaml(64)))
+        (mismatch_dir / "mlperf_endpoints_log_summary.json").write_text(json.dumps(_SUMMARY))
+        (mismatch_dir / "mlperf_endpoints_log_detail.json").write_text("{}")
+        (mismatch_dir / "run_metadata.json").write_text("{}")
+        (mismatch_dir / "report.txt").write_text("report\n")
+        src = mismatch_dir / "src" / "vllm"
+        src.mkdir(parents=True)
+        (src / ".gitkeep").write_text("")
+        acc = mismatch_dir / "accuracy"
+        acc.mkdir()
+        (acc / "accuracy.txt").write_text("ok")
+        (acc / "accuracy_result.json").write_text(json.dumps(_ACCURACY))
         report = _check(root)
         assert _warnings(report, "point-filename-concurrency")
 
     def test_invalid_point_yaml_is_skipped(self, tmp_path):
-        """A point_*.yaml that fails validation does not crash the checker."""
+        """A point.yaml that fails validation does not crash the checker."""
         root = _build_submission(tmp_path)
-        bad_yaml = root / "pareto" / "test-sys" / "llama3-70b" / "points" / "point_99.yaml"
-        bad_yaml.write_text("{bad yaml [")
+        bad_yaml = root / "test-sys" / "llama3-70b" / "r99"
+        bad_yaml.mkdir()
+        (bad_yaml / "point.yaml").write_text("{bad yaml [")
+        (bad_yaml / "mlperf_endpoints_log_summary.json").write_text(json.dumps(_SUMMARY))
+        (bad_yaml / "mlperf_endpoints_log_detail.json").write_text("{}")
+        (bad_yaml / "run_metadata.json").write_text("{}")
+        (bad_yaml / "report.txt").write_text("report\n")
+        src = bad_yaml / "src" / "vllm"
+        src.mkdir(parents=True)
+        (src / ".gitkeep").write_text("")
+        acc = bad_yaml / "accuracy"
+        acc.mkdir()
+        (acc / "accuracy.txt").write_text("ok")
+        (acc / "accuracy_result.json").write_text(json.dumps(_ACCURACY))
         report = _check(root)
-        # Should produce a point-config-valid error for the bad file
         assert _errors(report, "point-config-valid")
 
     def test_region_computation_error(self, tmp_path):
         """region-computation error when compute_regions raises ValueError."""
-        # compute_regions only raises if M <= 32, but SystemDescription enforces M > 32.
-        # Patch compute_regions to simulate an unexpected ValueError.
-        (tmp_path / "systems").mkdir()
-        (tmp_path / "systems" / "test-sys.json").write_text(json.dumps(_SYSTEM_DESC))
-        pareto_sys = tmp_path / "pareto" / "test-sys"
-        pareto_sys.mkdir(parents=True)
-        (pareto_sys / "llama3-70b").mkdir()
+        sys_dir = tmp_path / "test-sys"
+        sys_dir.mkdir()
+        (sys_dir / "system_desc.json").write_text(json.dumps(_SYSTEM_DESC))
+        (sys_dir / "llama3-70b").mkdir()
         with patch(
             "submission_checker.checker.compute_regions",
             side_effect=ValueError("M must be > 32"),
         ):
             report = _check(tmp_path)
         assert _errors(report, "region-computation")
-
-    def test_run_filename_non_numeric_suffix_ignored(self, tmp_path):
-        """Filename parsing errors (non-numeric suffix) are silently ignored."""
-        root = _build_submission(tmp_path)
-        # point_abc.yaml — stem is "point_abc", int("abc") raises ValueError
-        # This tests the except (IndexError, ValueError): pass branch
-        bad_name = root / "pareto" / "test-sys" / "llama3-70b" / "points" / "point_abc.yaml"
-        bad_name.write_text(yaml.dump(_make_run_yaml(64)))
-        result_dir = root / "pareto" / "test-sys" / "llama3-70b" / "results" / "point_64"
-        result_dir.mkdir(parents=True, exist_ok=True)
-        (result_dir / "mlperf_endpoints_log_summary.json").write_text(json.dumps(_SUMMARY))
-        (result_dir / "mlperf_endpoints_log_detail.json").write_text("{}")
-        report = _check(root)
-        # No run-filename-concurrency warning — the ValueError was swallowed
-        assert not _warnings(report, "point-filename-concurrency")
diff --git a/tests/submission_checker/test_checks_aggregate.py b/tests/submission_checker/test_checks_aggregate.py
index 48e792a..0f2fbf6 100644
--- a/tests/submission_checker/test_checks_aggregate.py
+++ b/tests/submission_checker/test_checks_aggregate.py
@@ -435,21 +435,16 @@ def test_inconsistent_datasets(self, tmp_path):
     def test_model_name_mismatch(self, tmp_path):
         model_dir = tmp_path / "actual-name"
         model_dir.mkdir(exist_ok=True)
-        (model_dir / "points").mkdir(exist_ok=True)
-        (model_dir / "results").mkdir(exist_ok=True)
-        (model_dir / "accuracy").mkdir(exist_ok=True)
         s = _summary()
         ctx = ModelContext(
             system_id="test-sys",
             system_desc=_system_desc(benchmark_model="expected-name"),
             model_dir=model_dir,
             regions=_REGIONS,
-            points_dir=model_dir / "points",
-            accuracy_dir=model_dir / "accuracy",
             all_point_count=7,
             valid_points=[],
             loaded_points=[(_config(), s)],
-            accuracy_result=None,
+            accuracy_results=[],
         )
         assert any(
             r.rule == "config-consistency-model" and r.severity == Severity.WARNING
@@ -465,12 +460,16 @@ def test_empty_results_skips(self, tmp_path):
 @pytest.mark.unit
 class TestAccuracyGateValidator:
     def test_no_accuracy_result_skips(self, tmp_path):
-        ctx = _model_ctx(tmp_path, accuracy_result=None)
+        ctx = _model_ctx(tmp_path, accuracy_results=[])
+        assert not any(r.rule == "accuracy-gate" for r in ctx._check_results)
+
+    def test_all_none_skips(self, tmp_path):
+        ctx = _model_ctx(tmp_path, accuracy_results=[None, None])
         assert not any(r.rule == "accuracy-gate" for r in ctx._check_results)
 
     def test_passed_accuracy_gate(self, tmp_path):
         ar = AccuracyResult(metric="rouge1", score=0.45, quality_target=0.43, passed=True)
-        ctx = _model_ctx(tmp_path, accuracy_result=ar)
+        ctx = _model_ctx(tmp_path, accuracy_results=[ar])
         assert any(
             r.rule == "accuracy-gate" and r.severity == Severity.INFO for r in ctx._check_results
         )
@@ -480,7 +479,15 @@ def test_passed_accuracy_gate(self, tmp_path):
 
     def test_failed_accuracy_gate(self, tmp_path):
         ar = AccuracyResult(metric="rouge1", score=0.30, quality_target=0.43, passed=False)
-        ctx = _model_ctx(tmp_path, accuracy_result=ar)
+        ctx = _model_ctx(tmp_path, accuracy_results=[ar])
+        assert any(
+            r.rule == "accuracy-gate" and r.severity == Severity.ERROR for r in ctx._check_results
+        )
+
+    def test_partial_failure_fails_gate(self, tmp_path):
+        passed = AccuracyResult(metric="rouge1", score=0.45, quality_target=0.43, passed=True)
+        failed = AccuracyResult(metric="rouge1", score=0.30, quality_target=0.43, passed=False)
+        ctx = _model_ctx(tmp_path, accuracy_results=[passed, failed])
         assert any(
             r.rule == "accuracy-gate" and r.severity == Severity.ERROR for r in ctx._check_results
         )
diff --git a/tests/submission_checker/test_checks_structure.py b/tests/submission_checker/test_checks_structure.py
index a6365a6..4e9b310 100644
--- a/tests/submission_checker/test_checks_structure.py
+++ b/tests/submission_checker/test_checks_structure.py
@@ -7,91 +7,143 @@
 import pytest
 
 from submission_checker.models import (
-    Division,
     ModelDir,
+    RunDir,
     Severity,
-    SrcDir,
-    SubmissionDir,
-    SystemPareto,
 )
 
 from .conftest import _passed
 
 
-@pytest.mark.unit
-class TestSubmissionDir:
-    def test_missing_dir(self, tmp_path):
-        (tmp_path / "systems").mkdir()
-        # pareto/ intentionally absent
-        sd = SubmissionDir(root=tmp_path)
-        rules = {r.rule for r in sd._check_results if r.severity == Severity.ERROR}
-        assert "required-dir" in rules
-
-    def test_both_present(self, tmp_path):
-        (tmp_path / "systems").mkdir()
-        (tmp_path / "pareto").mkdir()
-        sd = SubmissionDir(root=tmp_path)
-        assert _passed(sd._check_results)
-
-    def test_computed_paths(self, tmp_path):
-        sd = SubmissionDir(root=tmp_path)
-        assert sd.systems_dir == tmp_path / "systems"
-        assert sd.pareto_dir == tmp_path / "pareto"
-
-
-@pytest.mark.unit
-class TestSystemPareto:
-    def test_missing_system_pareto(self, tmp_path):
-        sp = SystemPareto(pareto_dir=tmp_path, system_id="sys-x")
-        assert any(r.severity == Severity.ERROR for r in sp._check_results)
-
-    def test_present(self, tmp_path):
-        (tmp_path / "sys-x").mkdir()
-        sp = SystemPareto(pareto_dir=tmp_path, system_id="sys-x")
-        assert _passed(sp._check_results)
-
-    def test_system_dir_computed(self, tmp_path):
-        sp = SystemPareto(pareto_dir=tmp_path, system_id="sys-x")
-        assert sp.system_dir == tmp_path / "sys-x"
-
-
 @pytest.mark.unit
 class TestModelDir:
-    def test_missing_subdir(self, tmp_path):
+    def test_missing_sweep_summary(self, tmp_path):
         model_dir = tmp_path / "model"
         model_dir.mkdir()
-        (model_dir / "points").mkdir()
-        (model_dir / "results").mkdir()
-        # accuracy/ absent
-        md = ModelDir(root=model_dir, system_id="sys-x", benchmark_model="llama3-70b")
-        assert any(r.severity == Severity.ERROR for r in md._check_results)
+        (model_dir / "sweep_distributions.csv").write_text("data\n")
+        # sweep_summary.csv intentionally absent
+        md = ModelDir(root=model_dir, system_name="sys-x", model_name="llama3-70b")
+        rules = {r.rule for r in md._check_results if r.severity == Severity.ERROR}
+        assert "model-sweep-file" in rules
 
-    def test_all_present(self, tmp_path):
+    def test_missing_sweep_distributions(self, tmp_path):
         model_dir = tmp_path / "model"
         model_dir.mkdir()
-        for d in ("points", "results", "accuracy"):
-            (model_dir / d).mkdir()
-        md = ModelDir(root=model_dir, system_id="sys-x", benchmark_model="llama3-70b")
+        (model_dir / "sweep_summary.csv").write_text("data\n")
+        # sweep_distributions.csv intentionally absent
+        md = ModelDir(root=model_dir, system_name="sys-x", model_name="llama3-70b")
+        rules = {r.rule for r in md._check_results if r.severity == Severity.ERROR}
+        assert "model-sweep-file" in rules
+
+    def test_both_sweep_files_present(self, tmp_path):
+        model_dir = tmp_path / "model"
+        model_dir.mkdir()
+        (model_dir / "sweep_summary.csv").write_text("data\n")
+        (model_dir / "sweep_distributions.csv").write_text("data\n")
+        md = ModelDir(root=model_dir, system_name="sys-x", model_name="llama3-70b")
         assert _passed(md._check_results)
 
-    def test_computed_paths(self, tmp_path):
-        md = ModelDir(root=tmp_path, system_id="sys-x", benchmark_model="llama3-70b")
-        assert md.points_dir == tmp_path / "points"
-        assert md.results_dir == tmp_path / "results"
-        assert md.accuracy_dir == tmp_path / "accuracy"
+    def test_run_dirs_property_matches_r_pattern(self, tmp_path):
+        model_dir = tmp_path / "model"
+        model_dir.mkdir()
+        (model_dir / "r16").mkdir()
+        (model_dir / "r256").mkdir()
+        (model_dir / "docs").mkdir()         # should be excluded
+        (model_dir / "sweep_summary.csv").touch()  # file, not dir
+        md = ModelDir(root=model_dir, system_name="sys-x", model_name="llama3-70b")
+        names = [d.name for d in md.run_dirs]
+        assert "r16" in names
+        assert "r256" in names
+        assert "docs" not in names
+
+    def test_run_dirs_empty_when_none(self, tmp_path):
+        model_dir = tmp_path / "model"
+        model_dir.mkdir()
+        md = ModelDir(root=model_dir, system_name="sys-x", model_name="llama3-70b")
+        assert md.run_dirs == []
 
 
 @pytest.mark.unit
-class TestSrcDir:
-    def test_standardized_missing_src(self, tmp_path):
-        sd = SrcDir(root=tmp_path, division=Division.STANDARDIZED)
-        assert any(r.severity == Severity.ERROR for r in sd._check_results)
-
-    def test_standardized_src_present(self, tmp_path):
-        (tmp_path / "src").mkdir()
-        sd = SrcDir(root=tmp_path, division=Division.STANDARDIZED)
-        assert _passed(sd._check_results)
-
-    def test_non_standardized_skipped(self, tmp_path):
-        sd = SrcDir(root=tmp_path, division=Division.SERVICED)
-        assert sd._check_results == []
+class TestRunDir:
+    def _make_valid_run_dir(self, tmp_path) -> None:
+        """Create a fully valid r16/ run directory."""
+        run = tmp_path / "r16"
+        run.mkdir()
+        (run / "point.yaml").write_text("concurrency: 16\n")
+        acc = run / "accuracy"
+        acc.mkdir()
+        (acc / "accuracy.txt").write_text("ROUGE-1: 0.45\n")
+        (acc / "accuracy_result.json").write_text("{}")
+        (run / "mlperf_endpoints_log_summary.json").write_text("{}")
+        (run / "mlperf_endpoints_log_detail.json").write_text("{}")
+        (run / "run_metadata.json").write_text("{}")
+        (run / "report.txt").write_text("report\n")
+        src = run / "src" / "vllm"
+        src.mkdir(parents=True)
+        (src / ".gitkeep").write_text("")
+
+    def test_all_present_passes(self, tmp_path):
+        self._make_valid_run_dir(tmp_path)
+        rd = RunDir(root=tmp_path / "r16", system_name="sys-x", model_name="llama3-70b")
+        assert _passed(rd._check_results)
+
+    def test_missing_point_yaml(self, tmp_path):
+        self._make_valid_run_dir(tmp_path)
+        (tmp_path / "r16" / "point.yaml").unlink()
+        rd = RunDir(root=tmp_path / "r16", system_name="sys-x", model_name="llama3-70b")
+        rules = {r.rule for r in rd._check_results if r.severity == Severity.ERROR}
+        assert "run-point-config" in rules
+
+    def test_missing_accuracy_dir(self, tmp_path):
+        self._make_valid_run_dir(tmp_path)
+        import shutil
+        shutil.rmtree(tmp_path / "r16" / "accuracy")
+        rd = RunDir(root=tmp_path / "r16", system_name="sys-x", model_name="llama3-70b")
+        rules = {r.rule for r in rd._check_results if r.severity == Severity.ERROR}
+        assert "run-accuracy-dir" in rules
+
+    def test_missing_summary_log(self, tmp_path):
+        self._make_valid_run_dir(tmp_path)
+        (tmp_path / "r16" / "mlperf_endpoints_log_summary.json").unlink()
+        rd = RunDir(root=tmp_path / "r16", system_name="sys-x", model_name="llama3-70b")
+        rules = {r.rule for r in rd._check_results if r.severity == Severity.ERROR}
+        assert "run-result-files" in rules
+
+    def test_missing_detail_log(self, tmp_path):
+        self._make_valid_run_dir(tmp_path)
+        (tmp_path / "r16" / "mlperf_endpoints_log_detail.json").unlink()
+        rd = RunDir(root=tmp_path / "r16", system_name="sys-x", model_name="llama3-70b")
+        rules = {r.rule for r in rd._check_results if r.severity == Severity.ERROR}
+        assert "run-result-files" in rules
+
+    def test_missing_run_metadata(self, tmp_path):
+        self._make_valid_run_dir(tmp_path)
+        (tmp_path / "r16" / "run_metadata.json").unlink()
+        rd = RunDir(root=tmp_path / "r16", system_name="sys-x", model_name="llama3-70b")
+        rules = {r.rule for r in rd._check_results if r.severity == Severity.ERROR}
+        assert "run-metadata" in rules
+
+    def test_missing_report_txt(self, tmp_path):
+        self._make_valid_run_dir(tmp_path)
+        (tmp_path / "r16" / "report.txt").unlink()
+        rd = RunDir(root=tmp_path / "r16", system_name="sys-x", model_name="llama3-70b")
+        rules = {r.rule for r in rd._check_results if r.severity == Severity.ERROR}
+        assert "run-report" in rules
+
+    def test_missing_src_dir(self, tmp_path):
+        self._make_valid_run_dir(tmp_path)
+        import shutil
+        shutil.rmtree(tmp_path / "r16" / "src")
+        rd = RunDir(root=tmp_path / "r16", system_name="sys-x", model_name="llama3-70b")
+        rules = {r.rule for r in rd._check_results if r.severity == Severity.ERROR}
+        assert "run-src-dir" in rules
+
+    def test_src_dir_no_impl_subdirs(self, tmp_path):
+        self._make_valid_run_dir(tmp_path)
+        import shutil
+        shutil.rmtree(tmp_path / "r16" / "src")
+        (tmp_path / "r16" / "src").mkdir()
+        # src/ exists but has no implementation subdirectories
+        rd = RunDir(root=tmp_path / "r16", system_name="sys-x", model_name="llama3-70b")
+        rules = {r.rule for r in rd._check_results if r.severity == Severity.ERROR}
+        assert "run-src-dir" in rules