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

Chinese Characters Corruption After Space in Output of qwen2.5-vl-7b-instruct@8bit Model #99

Open
panlilu opened this issue Feb 16, 2025 · 2 comments

Comments

@panlilu
Copy link

panlilu commented Feb 16, 2025

Description:

When running the qwen2.5-vl-7b-instruct@8bit model on lm-studio, there is an issue with Chinese character corruption, especially after a space. The issue occurs occasionally in the output, where Chinese characters after spaces are corrupted, resulting in some characters being displayed incorrectly.

Steps to Reproduce:
1. Run the qwen2.5-vl-7b-instruct@8bit model in lm-studio.
2. Provide input with Chinese text, for example: “喜剧 动画 冒险”.
3. In the output, some characters appear as corrupted, e.g., 喜剧 �画 �险.

Expected Behavior:
The model should handle Chinese characters correctly without any corruption, even after spaces.

Actual Behavior:
Some Chinese characters, especially those after spaces, are corrupted in the output.

Screenshot:

Image

Environment Information:
• Model version: qwen2.5-vl-7b-instruct@8bit
• System: lm-studio
• OS version: Sequoia 15.1.1
• Installation Type : Mac Silicon
• Mac Details : M4 pro; 24GB

Additional Information:
Let me know if more logs or information are needed. Thanks for your support!

@downerxl
Copy link

Same question in qwen2.5-vl-7b-instruct@f16

@datouxia
Copy link

datouxia commented Mar 8, 2025

me too, With "mlx-community/Qwen2.5-VL-72B-Instruct-4bit" this model.

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

No branches or pull requests

3 participants