Tobi Lutke f96766cce8 Fix GRPO model loading to use SFT base first ... The GRPO adapter was trained on merged SFT weights, so loading it directly on the base model results in 0% score. Added --sft-model parameter to evals/run.py to load SFT first, then apply GRPO adapter. With correct loading: GRPO scores 89.7% (all 26 queries Excellent). Updated README with correct GRPO score and loading instructions. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>		2026-01-25 00:46:07 -05:00
..
.gitignore	Refactor evals into separate run and score scripts	2026-01-25 00:40:33 -05:00
queries.txt	Refactor evals into separate run and score scripts	2026-01-25 00:40:33 -05:00
run.py	Fix GRPO model loading to use SFT base first	2026-01-25 00:46:07 -05:00
score.py	Refactor evals into separate run and score scripts	2026-01-25 00:40:33 -05:00