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

Strange Characters in Output After Update 0.3.6 in LM Studio #295

Open
jzsmoreno opened this issue Jan 10, 2025 · 9 comments
Open

Strange Characters in Output After Update 0.3.6 in LM Studio #295

jzsmoreno opened this issue Jan 10, 2025 · 9 comments

Comments

@jzsmoreno
Copy link

After updating LM Studio to version 0.3.6, when loading certain language models (including Qwen and others) and asking a question, the output consists of strange, nonsensical characters instead of coherent text. This issue does not occur with all models, but it affects several of them. The problem did not occur prior to the 0.3.6 update. It persists even after restarting the application and trying different inputs.
Captura de pantalla 2025-01-10 121446

@yagil
Copy link
Member

yagil commented Jan 10, 2025

Hi @jzsmoreno can you please press ctrl + shift + R and share a screenshot of the whole app? Thanks

@jzsmoreno
Copy link
Author

jzsmoreno commented Jan 10, 2025

Hi @jzsmoreno can you please press ctrl + shift + R and share a screenshot of the whole app? Thanks

Hi @yagil, sure! Thanks for your reply. Here is the screenshot.
Captura de pantalla 2025-01-10 123649

@arhontes-netizen
Copy link

arhontes-netizen commented Jan 12, 2025

Experiencing the same thing with Qwen2.5-Coder-14B-Instruct-GGUF currently, little time to test others. Same kind of output.

Some work fine, like Llama-3.1-Storm-8B-GGUF.

@droilogic
Copy link

droilogic commented Jan 13, 2025

I have the same problem on a win11 system with CPU + iGPU and it seems that the default Vulkan llama.cpp runtime causes the problem. Switching to CPU llama.cpp runtime solves the issue.
EDIT:

Posting screenshots of runtime config:
LM Studio RT1
this config fails (this was the default config after 1st time installation)

LM Studio RT2
this config works.

@TheDelvanD
Copy link

TheDelvanD commented Jan 14, 2025

Screenshot 2025-01-15 162547
I'm having the same problem although most of the characters are all square. Uninstalling, removing everything from appdata, and removing everything from .cache/lm-studio, then reinstalling solves the issue for the first response but all subsequent responses will be the square characters again.

@arhontes-netizen
Copy link

I wonder if this is somehow connected to many models now failing to even load since 0.3.6, or if the only commonality is the timing. Both problems began after updating.

@yagil
Copy link
Member

yagil commented Jan 15, 2025

To new folks commenting @TheDelvanD @arhontes-netizen @droilogic , please also include a screenshot of Ctrl + Shift + R. Thank you

@droilogic
Copy link

To new folks commenting @TheDelvanD @arhontes-netizen @droilogic , please also include a screenshot of Ctrl + Shift + R. Thank you

Post updated with screenshots.

@TheDelvanD
Copy link

To new folks commenting @TheDelvanD @arhontes-netizen @droilogic , please also include a screenshot of Ctrl + Shift + R. Thank you

Updated my post with screenshot.

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

6 participants
@yagil @jzsmoreno @droilogic @arhontes-netizen @TheDelvanD and others