gpjt/fineweb-gpt2-tokens download history
gpjt/fineweb-gpt2-tokens is a dataset on the Hugging Face Hub. In the last 30 days it was downloaded 112 times (49 in the last 7 days), and 457 times in total. It ranks #105,037 among datasets by monthly downloads.
FineWeb GPT-2 Tokens This is all of the 10B sample of FineWeb, split into 99% training data, 1% validation data, each converted to a single uint16 tensor of GPT-2 tokens. The text for each record in the original dataset was tokenized using tiktoken, then they were all concatenated togethe
Models trained on fineweb-gpt2-tokens
44 models list it as training data.
- gpjt/1xrtx3090-moe-1 369 downloads in 30 days
- gpjt/8xa100m40-baseline 253 downloads in 30 days
- gpjt/8xa100m40-lore-9-new-seed-output-only 155 downloads in 30 days
- gpjt/8xa100m40-lore-8-new-seed-both-smart-init 151 downloads in 30 days
- gpjt/8xa100m40-lore-3-both-smart-init 145 downloads in 30 days
- gpjt/8xa100m40-lore-7-new-seed-input-only 145 downloads in 30 days
- gpjt/8xa100m40-lore-6-new-seed-baseline-disabled 141 downloads in 30 days
- gpjt/8xa100m40-lore-4-input-only 141 downloads in 30 days
- gpjt/8xa100m40-lore-5-output-only 140 downloads in 30 days
- gpjt/8xa100m40-lore-2-both-ordinary-init 139 downloads in 30 days
- gpjt/8xa100m40-lore-11-output-only-corrected-smart-init 128 downloads in 30 days
- gpjt/8xa100m40-lore-10-both-corrected-smart-init 121 downloads in 30 days
Open gpjt/fineweb-gpt2-tokens on Hugging Face
Sister project: Paper Pulse, the upvote history of every Hugging Face Daily Paper.