Add datacenter_compute_allocation template (multi-reasoner)#68

Merged

cafzal merged 20 commits into

mainfrom

worktree-agent-a2bcce56b6427b60d

May 18, 2026

Commits on May 13, 2026

Add datacenter_compute_allocation template (multi-reasoner, chain follow-up to energy_grid_planning)
cafzal
committed
Regenerate v1 README index for datacenter_compute_allocation
cafzal
committed
Align datacenter_compute_allocation with peer multi-reasoner layout
cafzal
committed
Tighten datacenter_compute_allocation runbook + README to match energy_grid_planning style
cafzal
committed
Simplify datacenter_compute_allocation: drop chain-mode, keep conceptual lineage
cafzal
committed
Bake operator mental models into datacenter_compute_allocation
cafzal
committed
Reconcile docs to actual run output (post under-provisioning penalty)
cafzal
committed
Paste-test reconciliation: runbook Step 2 concept count + date range
cafzal
committed
Refactor Stage 1 to per-workload utilization-probability classification
cafzal
committed
Expand GNN training data with temporal per-(workload, month) observations
cafzal
committed
Reconcile docs to v5 temporal-GNN run output
cafzal
committed
Switch Stage 4 MIP from HiGHS (900s wall) to Gurobi (60s, typically OPTIMAL)
cafzal
committed
Reconcile docs to Gurobi run: 10s OPTIMAL vs 900s HiGHS TIME_LIMIT
cafzal
committed
Strip solver-specific framing from docs; default to highs+120s matching peer templates
cafzal
committed
Regenerate v1 README index after datacenter_compute_allocation description change
cafzal
committed
Mark datacenter_compute_allocation private (matches GNN-templates convention until GNN GAs)
cafzal
committed
Add reasoner-technique tags (Binary Classification, PageRank, MIP, Scenario Analysis) to match peer convention
cafzal
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add datacenter_compute_allocation template (multi-reasoner)#68

Add datacenter_compute_allocation template (multi-reasoner)#68
cafzal merged 20 commits into
mainfrom
worktree-agent-a2bcce56b6427b60d

Commits on May 13, 2026

Add datacenter_compute_allocation template (multi-reasoner, chain follow-up to energy_grid_planning)

Regenerate v1 README index for datacenter_compute_allocation

Align datacenter_compute_allocation with peer multi-reasoner layout

Tighten datacenter_compute_allocation runbook + README to match energy_grid_planning style

Simplify datacenter_compute_allocation: drop chain-mode, keep conceptual lineage

Bake operator mental models into datacenter_compute_allocation

Reconcile docs to actual run output (post under-provisioning penalty)

Paste-test reconciliation: runbook Step 2 concept count + date range

Refactor Stage 1 to per-workload utilization-probability classification

Expand GNN training data with temporal per-(workload, month) observations

Reconcile docs to v5 temporal-GNN run output

Switch Stage 4 MIP from HiGHS (900s wall) to Gurobi (60s, typically OPTIMAL)

Reconcile docs to Gurobi run: 10s OPTIMAL vs 900s HiGHS TIME_LIMIT

Strip solver-specific framing from docs; default to highs+120s matching peer templates

Regenerate v1 README index after datacenter_compute_allocation description change

Mark datacenter_compute_allocation private (matches GNN-templates convention until GNN GAs)

Add reasoner-technique tags (Binary Classification, PageRank, MIP, Scenario Analysis) to match peer convention

Commits on May 18, 2026

Refactor to canonical flat shape; trim sweep to 24 cells; reconcile docs to live Gurobi run

Final polish: drop "3D-scenario" framing, update stale row counts

Regenerate v1 README index after datacenter_compute_allocation description change