fs90's picture
Gemma 2 tokenizer vocabulary in reviewable .spm text form
cd7a149 verified
Raw
History Blame Contribute Delete
1.13 kB
Gemma is provided under and subject to the Gemma Terms of Use found at ai.google.dev/gemma/terms
---
MODIFICATION NOTICE
`gemma2.spm` in this repository is a MODIFIED form of Google's Gemma 2
`tokenizer.model`. It is not the original file.
What was changed: the SentencePiece protocol-buffer model was converted, id by
id, into a plain-text format — one line per token id, holding that id's piece,
score and piece type. Nothing was added, removed, reordered or rounded: all
256,000 pieces keep their ids, their scores and their SentencePiece piece types,
and the conversion is checked in both directions before the file is written.
What was NOT changed: no vocabulary entry, score, piece type or id assignment
differs from Google's file in any way.
Source: Google's Gemma 2 `tokenizer.model`, MD5 f9e2445870ec741aa6346bbd75531bb4
---
Use of Gemma and of Model Derivatives is subject to the use restrictions in
Section 3.2 of the Gemma Terms of Use, reproduced in the LICENSE file in this
repository. Those restrictions apply to you, and you must pass them on to anyone
you distribute this file or any derivative of it to.