T5-small is a large language model released by Google, released under the Apache-2.0 license, with a claimed knowledge cutoff of 2019-10-23. Its weights, configuration, and tokenizer files have been cryptographically hashed and are queued for on-chain attestation by Weight Registry.
Description
Casts every NLP task -- translation, summarization, classification, question-answering -- as text-to-text generation via one encoder-decoder architecture. The 'small' checkpoint is 60M parameters, the smallest of five sizes released up to 11B, all sharing the same C4 (Colossal Clean Crawled Corpus) pretraining recipe.
Metadata
Hugging Face stats
24,227,473 downloads on Hugging Face · 639 likes · pipeline: translation · library: transformers
Weight Hashes
Weights file
SHA-256bd944e5f1b3ad9b70dd9d00010a517059e19265671076b8b0a4a58d9491842bc
SHA3-256762f6d214a507e28398d76a6c51699a152fb4914df157cc9939464ccd385bcb1
BLAKE38ec9044cd73290ad0b6a29d5459845c36c078ef928bb27cf222ddc9497354867
Config file
SHA-256530e25060e3a8d5f7b0fcf53bfea9f3601165161f3b1914676e98d97cf07bcf1
SHA3-256595d34e46ace9c4e829b8631f01d27c29f0b968f1f9dd00b9f0c4777c8318d3c
BLAKE348b11a2887c203f2ae2b0bd67fe633820a67d1536d72675a1569973ef0cac7e3
Tokenizer file
SHA-256d2acde0d8d71dd30a711834b07781b9c89feaac33fd332f60507699282740066
SHA3-256f59741b250e9b839620d2a976a2dd1f64b2d458c4375fd792baf4900effdd30b
BLAKE39ce04442387ff177fd52c14e1bd1901ba563d71e956ca2cffd39eeb3ef4bb820
Verify it yourself
curl -L -o "model.safetensors" "https://huggingface.co/google-t5/t5-small/resolve/df1b051c49625cf57a3d0d8d3863ed4d13564fe4/model.safetensors" sha256sum "model.safetensors"
Expected SHA-256: bd944e5f1b3ad9b70dd9d00010a517059e19265671076b8b0a4a58d9491842bc