wassnameandClaudypoo 2001f41e03 refactor(c): extract eval.py (ref_logprobs, ablate_quarantine, eval_hack_solve) ...

Relocate the three read-only model helpers out of train.py into eval.py. They use
only torch + per_token_logps (proj) + compute_reward (rewards); no train globals.
Training numbers identical across all 4 smoke arms (resid/qE diagnostic cosines
show last-digit bf16 noise only). MODE_CODE stays in train.py.

Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>

2026-06-01 09:22:30 +00:00

..

projected_grpo

refactor(c): extract eval.py (ref_logprobs, ablate_quarantine, eval_hack_solve)

2026-06-01 09:22:30 +00:00