Pythia-160M is a large language model released by EleutherAI, released under the Apache-2.0 license, with a claimed knowledge cutoff of 2023-02-13. Its weights, configuration, and tokenizer files have been cryptographically hashed and are queued for on-chain attestation by Weight Registry.
Description
160M-parameter member of EleutherAI's Pythia suite -- 8 model sizes from 70M to 12B, each trained on the identical Pile dataset in the identical order, with 154 intermediate checkpoints saved per model -- built as a controlled experiment platform for interpretability and training-dynamics research.
Metadata
Hugging Face stats
3,315,221 downloads on Hugging Face · 45 likes · pipeline: text-generation · library: transformers
Weight Hashes
Weights file
SHA-25629d2e457a664e41c12c735f20a36dc0956a665f614a54ce5db21a32e75965270
SHA3-256f2161335edb66650ee49772780c27abdd9ed0017a50261fd4b68361df5dc74de
BLAKE367b53299d11986676d5e0df65e4e0a6b84454508a7001f2bdb5658591050a14a
Config file
SHA-25676eb275107220e450d31258f792a2efcbee109d8b62ae0088260057dec06362f
SHA3-256bebc1cdf33e27b5544a7f94e7e1b22e8ab17bae966d1ece8241616fb9b2bd5a2
BLAKE314fb84b6ea59c0b4d82d35531354b075c468a7feff711315494bd2b126ae08f8
Tokenizer file
SHA-256c24618a1b3e6a38167beff1c72cffd126c3a66254347304b50547d12c5f25624
SHA3-256848b77cfe64befb2401c373c6b6e4c5540e7b95d79c5280f8587bf3e86eca810
BLAKE348450cf08e78d969e4b459635395e5280319b8452583ea3807fe9d01fecbfa23
Verify it yourself
curl -L -o "model.safetensors" "https://huggingface.co/EleutherAI/pythia-160m/resolve/50f5173d932e8e61f858120bcb800b97af589f46/model.safetensors" sha256sum "model.safetensors"
Expected SHA-256: 29d2e457a664e41c12c735f20a36dc0956a665f614a54ce5db21a32e75965270