An implementation of model parallel GPT-2 and GPT-3-style models using the mesh-tensorflow library.
Vulnerabilities have been found in tensorflow which are patched in the most recent version. This release updates the codebase to use the secure version of tensorflow.
This release also fixes a small but significant bug in how documents are loaded. For details, see #230
We're proud to release two pretrained GPT-Neo models trained on The Pile, the weights and configs can be freely downloaded from the-eye.eu.
1.3B: https://the-eye.eu/eleuther_staging/gptneo-release/GPT3_XL/
2.7B: https://the-eye.eu/eleuther_staging/gptneo-release/GPT3_2-7B/
For more information on how to get these set up, see the colab notebook, or read through the rest of the readme.
This repository will be (mostly) archived as we move focus to our GPU training repo, GPT-Neox