wassname b4e76525c1 Per-prompt grouping, hint default, ratio diagnostic, LR=3e-4 ...

- load_problems applies the simple_overwrite_tests hint by default (matches
  ariahw's load-time hint registry). Both pools now see the identical prompt.
- Pool files keyed by prompt_id (prompt_NNNN.jsonl.gz); each = G rollouts of
  one problem. Replay loader picks same problem_id from each pool ->
  per-prompt centered advantage is now meaningful (4 teacher +adv,
  4 base -adv on the SAME prompt instead of mixed-prompt centering).
- Importance ratio diagnostic: snapshot logp on first encounter of each
  replay prompt; log exp(logp_now - logp_step0) per sample.
  Healthy ~2-5; explosion >10 == overfit on teacher tokens.
- Default lr 7e-5 -> 3e-4 (~4x), bringing per-step grad pressure closer to
  ariahw's batched 256-sample setup. Grad-clip=1 still protects.

2026-05-25 22:03:50 +00:00

..

projected_grpo

Per-prompt grouping, hint default, ratio diagnostic, LR=3e-4

2026-05-25 22:03:50 +00:00