recursal/DCLM-10B-Qwen2-binidx download history

recursal/DCLM-10B-Qwen2-binidx is a text generation dataset on the Hugging Face Hub. In the last 30 days it was downloaded 44 times (9 in the last 7 days), and 1,035 times in total. It ranks #203,866 among datasets by monthly downloads.

This repository contains the DCLM-10B-Qwen2-binidx dataset, a large-scale text corpus used as training data for the RADLADS: Rapid Attention Distillation to Linear Attention Decoders at Scale protocol. RADLADS proposes a method for rapidly converting traditional softmax attention transformers into e

Open recursal/DCLM-10B-Qwen2-binidx on Hugging Face

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