Experiment: Decrease WARMDOWN_RATIO from 0.825 to 0.75 to allow more time at higher learning rates before the final LR floor.

Campaign

• Campaign: schedule-tuning

Hypothesis

Decrease WARMDOWN_RATIO from 0.825 to 0.75 to allow more time at higher learning rates before the final LR floor.

Parent Context

• Parent master hash: 935fdbf9f4ae8a5ef5bcb76552acea2bc5801965
• Master val_bpb at dispatch: 0.962777
• Worker id: worker-1
• Worktree: /Users/ben/code/open-autolab/.runtime/worktrees/warmdown-075

Single Variable

WARMDOWN_RATIO = 0.75 (was 0.825)

Expected Upside

More time at higher learning rates before the final LR floor may improve convergence quality.

Duplicate Check

No prior experiments have tested WARMDOWN_RATIO changes against this master.

Runtime

• Log path: /Users/ben/code/open-autolab/research/live/warmdown-075.log
• Launcher: uv run scripts/opencode_worker.py run warmdown-075

Allowed Edit Scope

• train.py only

Run Plan

• Refresh master with uv run scripts/refresh_master.py --fetch-dag
• Run uv run scripts/hf_job.py preflight
• Run uv run scripts/hf_job.py launch --mode experiment
• Stream logs to the reserved path
• Parse uv run scripts/parse_metric.py /Users/ben/code/open-autolab/research/live/warmdown-075.log

Result

• HF Job: 69cc194534fa24114ddf48bd
• val_bpb: 0.962980
• Master val_bpb: 0.962777
• Delta: +0.000203 (degraded)
• training_seconds: 300.1
• total_seconds: 431.5
• peak_vram_mb: 33609.6
• mfu_percent: 43.88
• total_tokens_M: 304.7
• num_steps: 2325
• Submitted: no
• Interpretation: Decreasing WARMDOWN_RATIO from 0.825 to 0.75 slightly degraded validation bpb. The longer warmdown phase in master appears beneficial for final convergence.
• Failure mode: Slight regression (+0.000203 bpb).

Memory-Keeper Handoff

• One short note for research/notes.md: Added experiment entry with FAILED status.
• Any do-not-repeat update: Added to Schedule Tuning section.