the model(student) after finetuning outperforms 410m-deduped model on wsc accuracy.
lm-eval : aloobun/distill-p-test
lm-eval : EleutherAI/pythia-410m-deduped
Files info