You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This repo is primarily intended to be used for training. Our expectation is that people will generally want to modify/tweak the training scripts before running them, so I don't think packaging them makes sense.
For inference, you can use the original tokenizers library. (You don't need this repo.)
Additionally, the tokenizers SuperBPE fork can be installed via its git address via pip or any other python package manager.
Activity
PythonNut commentedon Jan 17, 2026
This repo is primarily intended to be used for training. Our expectation is that people will generally want to modify/tweak the training scripts before running them, so I don't think packaging them makes sense.
For inference, you can use the original tokenizers library. (You don't need this repo.)
Additionally, the tokenizers SuperBPE fork can be installed via its git address via pip or any other python package manager.