sfanm/gsm8k-boxed-b16r1k-val-dumps download history

sfanm/gsm8k-boxed-b16r1k-val-dumps is a dataset on the Hugging Face Hub. In the last 30 days it was downloaded 23 times (9 in the last 7 days), and 82 times in total. It ranks #338,096 among datasets by monthly downloads.

gsm8k-boxed-b16r1k — validation generations Validation rollout dumps for the b16r1k cohort: SmolLM2-360M on GSM8K with the boxed-answer recipe, 16 prompts x group 16 (256 seqs/step), 1,024-token response budget, REQUIRE_EOS reward (train-only), 3,000 training steps. Companion wandb pr

Open sfanm/gsm8k-boxed-b16r1k-val-dumps on Hugging Face

Sister project: Paper Pulse, the upvote history of every Hugging Face Daily Paper.