[GLM-4.7] GLM-4.7-Flash Model Supoort #43031
Conversation
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
ArthurZucker
left a comment
ArthurZucker
left a comment
There was a problem hiding this comment.
Very nice! Almost nothing to say but that we might want to go explicit for mlp type maybe? not a must its more for use @Cyrilvallez and @vasqu ?
src/transformers/models/glm4_moe_lite/configuration_glm4_moe_lite.py
Outdated
Show resolved
Hide resolved
|
[For maintainers] Suggested jobs to run (before merge) run-slow: auto, glm4_moe, glm4_moe_lite |
| - local: model_doc/xlstm | ||
| title: xLSTM | ||
| - local: model_doc/glm4_moe_lite | ||
| title: y |
There was a problem hiding this comment.
This seems incorrect, could we replace it with e.g. glm4_moe_lite
There was a problem hiding this comment.
Is this urgent to fix? I can wait and provide a new PR with the model introduction and these changes after I've completed the model description.
There was a problem hiding this comment.
Not urgent, just something that should be fixed at some point
zRzRzRzRzRzRzR
changed the title
5 participants
GLM-Lite is the class updated for GLM-4.7-Flash, which adopts a different model architecture from GLM-4.7. The intermediate state was named GLM-Lite and has now been renamed to GLM-4.7-Flash. At the same time, a PR has been submitted to improve the documentation.