kswhitecross/RecaLLM-sft download history

kswhitecross/RecaLLM-sft is a dataset on the Hugging Face Hub. In the last 30 days it was downloaded 21 times (1 in the last 7 days), and 216 times in total. It ranks #361,218 among datasets by monthly downloads.

RecaLLM SFT Training Data Supervised finetuning data for RecaLLM. Contains 1,795 reasoning traces annotated with verbatim recall spans, used as a cold start before GRPO reinforcement learning. How the data was created Six teacher models (Llama-70B-R1, Llama-8B-R1, Qwen

Open kswhitecross/RecaLLM-sft on Hugging Face

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