yavuz-ai/self-reward-collapse-terse download history

yavuz-ai/self-reward-collapse-terse is a text generation dataset on the Hugging Face Hub. In the last 30 days it was downloaded 86 times (12 in the last 7 days), and 293 times in total. It ranks #127,866 among datasets by monthly downloads.

self-reward-collapse-terse Per-round trajectory and held-out samples from an iterative self-training loop on GSM8K (Qwen2.5-7B-Instruct, LoRA DPO, 6 rounds). Same loop every arm; only the preference-label source differs. This dataset is the terse arm: pairs labelled by a deliberately

Open yavuz-ai/self-reward-collapse-terse on Hugging Face

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