github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #2211] Mistral v0.2 hangs after repeatedly writing same token #1265

New Issue
Closed
opened 2026-04-12 11:03:35 -05:00 by GiteaMirror · 6 comments
GiteaMirror commented 2026-04-12 11:03:35 -05:00
Owner
Copy Link

Originally created by @arch-user-france1 on GitHub (Jan 26, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/2211

grafik 
➜  ~ ollama ls              
NAME            ID              SIZE    MODIFIED   
mistral:v0.2    61e88e884507    4.1 GB  2 days ago

Crashed, the ollama runner using 300 MB of RAM without active CPU or GPU usage.
Running on MPS. Conversation can not be continued anymore, ollama's text generation API hangs:

Looking forward to our interaction! Let me know if you have any specific requests or questions. Cheers! 🤘🏼💻💪🏼🌠🚀✨🔬🧩👨‍💻👩‍💻🐶😸🌹🏛️🌅🌄🌈🌊🏖️🚴🏼‍♂️🚶🏼‍♀️🚗🚧🚢🚤🚣🏻‍♂️🚣🏻‍♀️🛳️🚤🚢🛵🏋️🎾🤽🏻‍♂️🤸🏾‍♀️🧞🏽🧜🏼‍♂️🧜🏼‍♀️🦇🐍🕷️🕸️🌼🌱🌺🍎🍌🍊🍌🤘🏼💻💪🏼🌠🚀✨🔬🧩👨‍💻👩‍💻🐶😸🌹🏛️🌅🌄🌈🌊🏖️🚴🏼‍♂️🚶🏼‍♀️🚗🚧🚢🚤🚣🏻‍♂️🚣🏻‍♀️🛳️🚤🚢🛵🏋️🎾🤽🏻‍♂️🤸🏾‍♀️🧞🏽🧜🏼‍♂️🧜🏼‍♀️🦇🐍🕷️🕸️🌼🌱🌺🍎🍌🍊🍌🍇🥝🤴🏻🤴🏼👸🏽👸🏼🦄🐲🐉🐉🦋🐘🐘🧁🧁🎂🍰🥧🧁🧀🧀🌭🌮🍔🍟🍕🍖🍲🍱🥙🏴󠁧󠁢󠁥󠁮󠁧󠁿🇩🇪🇫🇷🇯🇵🇹🇼🇨🇭🇦🇹🇱🇻🇵🇸🇪🇸🇮🇹🇬🇧🇯🇴🇩🇿🇨🇼🇪🇪🇱🇾🇲🇪🇷🇸🇭🇷🇹🇹🇮🇸🇬🇧🇵🇺🇿🇦🇫🇷🇨🇩🇳🇱🇭🇷🇭🇾🇳🇴🇩🇰🇪🇸🇮🇸🇫🇷🇩🇰🇧🇪🇬🇧🇯🇵🇦🇺🇨
🇭🇩🇿🇱🇹🇲🇦🇷🇸🇫🇷🇧🇬🇮🇳🇪🇸🇭🇷🇲🇨🇳🇱🇯🇵🇩🇿🇩🇰🇪🇸🇫🇷🇦🇺🇹🇼🇧🇬🇭🇳🇱🇮🇶🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🥝🤴🏻🤴🏼👸🏽👸🏼🦄🐲🐉🐉🦋🐘🐘🧁🧁🎂🍰🥧🧁🧀🧀🌭🌮🍔🍟🍕🍖🍲🍱🥙🏴󠁧󠁢󠁥󠁮󠁧󠁿🇩🇪🇫🇷🇯🇵🇹🇼🇨🇭🇦🇹🇱🇻🇵🇸🇪🇸🇮🇹🇬🇧🇯🇴🇩🇿🇨🇼🇪🇪🇱🇾🇲🇪🇷🇸🇭🇷🇹🇹🇮🇸🇬🇧🇵🇺🇿🇦🇫🇷🇨🇩🇳🇱🇭🇷🇭🇾🇳🇴🇩🇰🇪🇸🇮🇸🇫🇷🇩🇰🇧🇪🇬🇧🇯🇵🇦🇺🇨🇭🇩🇿🇱🇹🇲🇦🇷🇸🇫🇷🇧🇬🇮🇳🇪🇸🇭🇷🇲🇨🇳🇱🇯🇵🇩🇿🇩🇰🇪🇸🇫🇷🇦🇺🇹🇼🇧🇬🇭🇳🇱🇮🇶🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹����🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹^C

>>> General tech news and trends
⠹

New conversations can not be started. Bug may only be reproduced with the correct seed, as mistral usually does not have any temperature problems.

Originally created by @arch-user-france1 on GitHub (Jan 26, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/2211 <img width="1446" alt="grafik" src="https://github.com/ollama/ollama/assets/72965843/daff8519-4262-46f1-b52d-d11b246355b4"> ```bash ➜ ~ ollama ls NAME ID SIZE MODIFIED mistral:v0.2 61e88e884507 4.1 GB 2 days ago ``` Crashed, the ollama runner using 300 MB of RAM without active CPU or GPU usage. Running on MPS. Conversation can not be continued anymore, ollama's text generation API hangs: ```bash Looking forward to our interaction! Let me know if you have any specific requests or questions. Cheers! 🤘🏼💻💪🏼🌠🚀✨🔬🧩👨‍💻👩‍💻🐶😸🌹🏛️🌅🌄🌈🌊🏖️🚴🏼‍♂️🚶🏼‍♀️🚗🚧🚢🚤🚣🏻‍♂️🚣🏻‍♀️🛳️🚤🚢🛵🏋️🎾🤽🏻‍♂️🤸🏾‍♀️🧞🏽🧜🏼‍♂️🧜🏼‍♀️🦇🐍🕷️🕸️🌼🌱🌺🍎🍌🍊🍌🤘🏼💻💪🏼🌠🚀✨🔬🧩👨‍💻👩‍💻🐶😸🌹🏛️🌅🌄🌈🌊🏖️🚴🏼‍♂️🚶🏼‍♀️🚗🚧🚢🚤🚣🏻‍♂️🚣🏻‍♀️🛳️🚤🚢🛵🏋️🎾🤽🏻‍♂️🤸🏾‍♀️🧞🏽🧜🏼‍♂️🧜🏼‍♀️🦇🐍🕷️🕸️🌼🌱🌺🍎🍌🍊🍌🍇🥝🤴🏻🤴🏼👸🏽👸🏼🦄🐲🐉🐉🦋🐘🐘🧁🧁🎂🍰🥧🧁🧀🧀🌭🌮🍔🍟🍕🍖🍲🍱🥙🏴󠁧󠁢󠁥󠁮󠁧󠁿🇩🇪🇫🇷🇯🇵🇹🇼🇨🇭🇦🇹🇱🇻🇵🇸🇪🇸🇮🇹🇬🇧🇯🇴🇩🇿🇨🇼🇪🇪🇱🇾🇲🇪🇷🇸🇭🇷🇹🇹🇮🇸🇬🇧🇵🇺🇿🇦🇫🇷🇨🇩🇳🇱🇭🇷🇭🇾🇳🇴🇩🇰🇪🇸🇮🇸🇫🇷🇩🇰🇧🇪🇬🇧🇯🇵🇦🇺🇨 🇭🇩🇿🇱🇹🇲🇦🇷🇸🇫🇷🇧🇬🇮🇳🇪🇸🇭🇷🇲🇨🇳🇱🇯🇵🇩🇿🇩🇰🇪🇸🇫🇷🇦🇺🇹🇼🇧🇬🇭🇳🇱🇮🇶🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🥝🤴🏻🤴🏼👸🏽👸🏼🦄🐲🐉🐉🦋🐘🐘🧁🧁🎂🍰🥧🧁🧀🧀🌭🌮🍔🍟🍕🍖🍲🍱🥙🏴󠁧󠁢󠁥󠁮󠁧󠁿🇩🇪🇫🇷🇯🇵🇹🇼🇨🇭🇦🇹🇱🇻🇵🇸🇪🇸🇮🇹🇬🇧🇯🇴🇩🇿🇨🇼🇪🇪🇱🇾🇲🇪🇷🇸🇭🇷🇹🇹🇮🇸🇬🇧🇵🇺🇿🇦🇫🇷🇨🇩🇳🇱🇭🇷🇭🇾🇳🇴🇩🇰🇪🇸🇮🇸🇫🇷🇩🇰🇧🇪🇬🇧🇯🇵🇦🇺🇨🇭🇩🇿🇱🇹🇲🇦🇷🇸🇫🇷🇧🇬🇮🇳🇪🇸🇭🇷🇲🇨🇳🇱🇯🇵🇩🇿🇩🇰🇪🇸🇫🇷🇦🇺🇹🇼🇧🇬🇭🇳🇱🇮🇶🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹����🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹🇹^C >>> General tech news and trends ⠹ ``` New conversations can not be started. Bug may only be reproduced with the correct seed, as mistral usually does not have any temperature problems.
GiteaMirror added the bug label 2026-04-12 11:03:35 -05:00
GiteaMirror commented 2026-04-12 11:03:36 -05:00
Author
Owner
Copy Link

@jukofyork commented on GitHub (Jan 28, 2024):

Make sure you use the exact promp format from the huggingface repository tokenizer. Mistral and Mixtral are super picky about the prompt format and just adding an extra space can make them go crazy (IIRC the default template from the ollama model download page adds a newline after the prompt that shouldn't be there):

https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2

<s>[INST] {{ .Prompt}} [/INST]

Adding a space after the ']' serverly degrades the model from my testing.

<!-- gh-comment-id:1913713321 --> @jukofyork commented on GitHub (Jan 28, 2024): Make sure you use the exact promp format from the huggingface repository tokenizer. Mistral and Mixtral are super picky about the prompt format and just adding an extra space can make them go crazy (IIRC the default template from the ollama model download page adds a newline after the prompt that shouldn't be there): https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2 `<s>[INST] {{ .Prompt}} [/INST]` Adding a space after the ']' serverly degrades the model from my testing.
GiteaMirror commented 2026-04-12 11:03:37 -05:00
Author
Owner
Copy Link

@arch-user-france1 commented on GitHub (Jan 28, 2024):

I did not modify the prompt format.

<!-- gh-comment-id:1913713854 --> @arch-user-france1 commented on GitHub (Jan 28, 2024): I did not modify the prompt format.
GiteaMirror commented 2026-04-12 11:03:38 -05:00
Author
Owner
Copy Link

@jukofyork commented on GitHub (Jan 28, 2024):

I did not modify the prompt format.

See my edit above.

<!-- gh-comment-id:1913714796 --> @jukofyork commented on GitHub (Jan 28, 2024): > I did not modify the prompt format. See my edit above.
GiteaMirror commented 2026-04-12 11:03:39 -05:00
Author
Owner
Copy Link

@jukofyork commented on GitHub (Jan 28, 2024):

It won't fix the crash though so apologies if that's what you meant :)

It's definitely worth double checking every template you use on huggingface IMO as a lot of the local LLMs are severely hurt by them and nearly all on the official ollama download page are slightly wrong. I stopped 99% of the problems I was getting with weird repeating gibberish and Chinese Unicode characters now.

<!-- gh-comment-id:1913715586 --> @jukofyork commented on GitHub (Jan 28, 2024): It won't fix the crash though so apologies if that's what you meant :) It's definitely worth double checking every template you use on huggingface IMO as a lot of the local LLMs are severely hurt by them and nearly all on the official ollama download page are slightly wrong. I stopped 99% of the problems I was getting with weird repeating gibberish and Chinese Unicode characters now.
GiteaMirror commented 2026-04-12 11:03:39 -05:00
Author
Owner
Copy Link

@arch-user-france1 commented on GitHub (Jan 28, 2024):

Thank you for your help. Yeah, I haven't modified the prompt format of Ollama, and I indeed reported the bug because of the crash, and not because of the problem with the model. Still, thank you for your suggestions.

<!-- gh-comment-id:1913718063 --> @arch-user-france1 commented on GitHub (Jan 28, 2024): Thank you for your help. Yeah, I haven't modified the prompt format of Ollama, and I indeed reported the bug because of the crash, and not because of the problem with the model. Still, thank you for your suggestions.
GiteaMirror commented 2026-04-12 11:03:40 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Mar 12, 2024):

This was fixed a while ago, but we lost track of the issue. I'll go ahead and close it out.

<!-- gh-comment-id:1992705016 --> @pdevine commented on GitHub (Mar 12, 2024): This was fixed a while ago, but we lost track of the issue. I'll go ahead and close it out.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#1265
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?