leejason2025/magic-cap1b-data download history

leejason2025/magic-cap1b-data is a dataset on the Hugging Face Hub. In the last 30 days it was downloaded 18 times (5 in the last 7 days), and 72 times in total. It ranks #402,156 among datasets by monthly downloads.

magic-cap1b training data ~1B tokens, 1,953,123 blocks of 512 tokens (uint16 ids, vocab 276): 50% FineWeb-Edu sample-10BT (byte-level, digits atomic) + 50% synthetic arithmetic in 4 evidence families (direct_sp scratchpad / verbal / code with computed values / primitive). Widened spec

Open leejason2025/magic-cap1b-data on Hugging Face

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