From 7a567efb9b88674a60702047210c1f393487d4a4 Mon Sep 17 00:00:00 2001
From: danielbusnz-lgtm <danielbusnz@gmail.com>
Date: Fri, 29 May 2026 13:13:38 -0400
Subject: [PATCH] Version the training data with DVC

Track data/*.jsonl with DVC (git holds the .dvc pointers, data lives in an
R2 remote). The frozen evals/holdout.jsonl stays committed directly so the
benchmark reproduces without DVC. Credentials live in the gitignored
.dvc/config.local, only the endpoint is committed.
---
 .dvc/.gitignore            |  3 +++
 .dvc/config                |  5 +++++
 .dvcignore                 |  3 +++
 README.md                  | 12 ++++++++++++
 data/agent.jsonl.dvc       |  5 +++++
 data/augmented.jsonl.dvc   |  5 +++++
 data/chat.jsonl.dvc        |  5 +++++
 data/find_action.jsonl.dvc |  5 +++++
 data/integration.jsonl.dvc |  5 +++++
 data/memory.jsonl.dvc      |  5 +++++
 10 files changed, 53 insertions(+)
 create mode 100644 .dvc/.gitignore
 create mode 100644 .dvc/config
 create mode 100644 .dvcignore
 create mode 100644 data/agent.jsonl.dvc
 create mode 100644 data/augmented.jsonl.dvc
 create mode 100644 data/chat.jsonl.dvc
 create mode 100644 data/find_action.jsonl.dvc
 create mode 100644 data/integration.jsonl.dvc
 create mode 100644 data/memory.jsonl.dvc

diff --git a/.dvc/.gitignore b/.dvc/.gitignore
new file mode 100644
index 0000000..528f30c
--- /dev/null
+++ b/.dvc/.gitignore
@@ -0,0 +1,3 @@
+/config.local
+/tmp
+/cache
diff --git a/.dvc/config b/.dvc/config
new file mode 100644
index 0000000..6692a99
--- /dev/null
+++ b/.dvc/config
@@ -0,0 +1,5 @@
+[core]
+    remote = routelet-r2
+['remote "routelet-r2"']
+    url = s3://aegis-routelet-samples/dvc
+    endpointurl = https://3888bcbf92cc98efac814d0108cf9358.r2.cloudflarestorage.com
diff --git a/.dvcignore b/.dvcignore
new file mode 100644
index 0000000..5197305
--- /dev/null
+++ b/.dvcignore
@@ -0,0 +1,3 @@
+# Add patterns of files dvc should ignore, which could improve
+# the performance. Learn more at
+# https://dvc.org/doc/user-guide/dvcignore
diff --git a/README.md b/README.md
index 6c11059..d590dec 100644
--- a/README.md
+++ b/README.md
@@ -59,6 +59,18 @@ One JSON object per line:
 {"text": "play despacito on spotify", "intent": "integration"}
 ```
 
+## Data versioning
+
+The training pool under `data/` is versioned with [DVC](https://dvc.org); git
+tracks the `*.jsonl.dvc` pointers, the data lives in an R2 remote. After cloning:
+
+```
+dvc pull   # needs R2 credentials in .dvc/config.local (see dvc remote modify --local)
+```
+
+The frozen eval set, `evals/holdout.jsonl`, is committed to git directly so the
+benchmark is reproducible without DVC.
+
 ## License
 
 MIT
diff --git a/data/agent.jsonl.dvc b/data/agent.jsonl.dvc
new file mode 100644
index 0000000..289bf7a
--- /dev/null
+++ b/data/agent.jsonl.dvc
@@ -0,0 +1,5 @@
+outs:
+- md5: befb0c5056c727158bb34a9032ab5a27
+  size: 5835
+  hash: md5
+  path: agent.jsonl
diff --git a/data/augmented.jsonl.dvc b/data/augmented.jsonl.dvc
new file mode 100644
index 0000000..1c0f402
--- /dev/null
+++ b/data/augmented.jsonl.dvc
@@ -0,0 +1,5 @@
+outs:
+- md5: 3125afd6cc9be7bd14b1442d965baf3f
+  size: 95998
+  hash: md5
+  path: augmented.jsonl
diff --git a/data/chat.jsonl.dvc b/data/chat.jsonl.dvc
new file mode 100644
index 0000000..cd317a2
--- /dev/null
+++ b/data/chat.jsonl.dvc
@@ -0,0 +1,5 @@
+outs:
+- md5: e2e66e5449bd901c87f00ca6ef8096eb
+  size: 7968
+  hash: md5
+  path: chat.jsonl
diff --git a/data/find_action.jsonl.dvc b/data/find_action.jsonl.dvc
new file mode 100644
index 0000000..637a877
--- /dev/null
+++ b/data/find_action.jsonl.dvc
@@ -0,0 +1,5 @@
+outs:
+- md5: 14f4fae92a55c1222601a1523e423ee2
+  size: 5225
+  hash: md5
+  path: find_action.jsonl
diff --git a/data/integration.jsonl.dvc b/data/integration.jsonl.dvc
new file mode 100644
index 0000000..6149f97
--- /dev/null
+++ b/data/integration.jsonl.dvc
@@ -0,0 +1,5 @@
+outs:
+- md5: 778f1024b240d116989879ebf14166fd
+  size: 15150
+  hash: md5
+  path: integration.jsonl
diff --git a/data/memory.jsonl.dvc b/data/memory.jsonl.dvc
new file mode 100644
index 0000000..4827ba4
--- /dev/null
+++ b/data/memory.jsonl.dvc
@@ -0,0 +1,5 @@
+outs:
+- md5: f7a7d0cd75626226c54387ca480d48b9
+  size: 3912
+  hash: md5
+  path: memory.jsonl