pngwn/memcap-random-data download history

pngwn/memcap-random-data is a dataset on the Hugging Face Hub. In the last 30 days it was downloaded 30 times (12 in the last 7 days), and 186 times in total. It ranks #383,577 among datasets by monthly downloads.

memcap-random-data Uniformly random token datasets for measuring the memorization capacity of small language models, reproducing the protocol of "How much do language models memorize?" (Sec. 3.2) at tiny scale. Because the data has no structure, generalization is impossible: every bit

Open pngwn/memcap-random-data on Hugging Face

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