Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Megrez] supported #67

Merged
merged 2 commits into from
Dec 27, 2024
Merged

[Megrez] supported #67

merged 2 commits into from
Dec 27, 2024

Conversation

wlc952
Copy link
Contributor

@wlc952 wlc952 commented Dec 27, 2024

No description provided.

Copy link
Collaborator

@fangz-ai fangz-ai left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

export_onnx.py

  1. ID_EOS = [120005, 120000, 120025]最好不要定死,一般会有tokenizer.im_eos_id这种
  2. 建议最好还是不要支持lmhead+top1这种贪婪采样,因为小模型,3B还好1.8B这种可能容易出现重复token的情况。建议同时支持贪婪和惩罚

chat.cpp
NUM_LAYERS最好也不要写死,直接写成32,不然以后不太好改,可以读模型config,json
或者放在pipeline.py里面最为一个--num_layers的参数

其他都写得蛮好的

@fangz-ai fangz-ai merged commit 2878c1a into sophgo:main Dec 27, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants