[GH-ISSUE #3668] wizardlm2:8x22b-q4_0 is failing to generate on M3 max #64296

Closed
opened 2026-05-03 16:58:35 -05:00 by GiteaMirror · 5 comments
Owner

Originally created by @xansrnitu on GitHub (Apr 16, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/3668

What is the issue?

I am trying to run wizardlm2:8x22b-q4_0 on M3 max machine. However, it is not generating meaningful token to the given prompt.

Screenshot 2024-04-16 at 11 35 33 AM

ollama version is 0.1.32

What did you expect to see?

No response

Steps to reproduce

No response

Are there any recent changes that introduced the issue?

No response

OS

macOS

Architecture

arm64

Platform

No response

Ollama version

0.1.32

GPU

Apple

GPU info

M3 Max 128GB

CPU

Apple

Other software

No response

Originally created by @xansrnitu on GitHub (Apr 16, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/3668 ### What is the issue? I am trying to run wizardlm2:8x22b-q4_0 on M3 max machine. However, it is not generating meaningful token to the given prompt. <img width="520" alt="Screenshot 2024-04-16 at 11 35 33 AM" src="https://github.com/ollama/ollama/assets/63858582/2ceeaff6-dfe2-4c8b-b63a-a33ac9ecb1d2"> ollama version is 0.1.32 ### What did you expect to see? _No response_ ### Steps to reproduce _No response_ ### Are there any recent changes that introduced the issue? _No response_ ### OS macOS ### Architecture arm64 ### Platform _No response_ ### Ollama version 0.1.32 ### GPU Apple ### GPU info M3 Max 128GB ### CPU Apple ### Other software _No response_
GiteaMirror added the bug label 2026-05-03 16:58:36 -05:00
Author
Owner

@adrienbrault commented on GitHub (Apr 16, 2024):

Hey, same on an M1 Max 64GB, with:

  • dbrx:132b-instruct-q2_K
  • wizardlm2:8x22b-q2_K
  • mixtral:8x22b-text-v0.1-q2_K

Note that command-r-plus:104b-q2_K seems to work fine.

$ ollama --version
ollama version is 0.1.32

$ # NOT OK

$ ollama run mixtral:8x22b-text-v0.1-q2_K 'What is Symfony UX?'
        "
$

$ ollama run wizardlm2:8x22b-q2_K 'What is Symfony UX?'
!$▅	 $▅▅#


▅"#$
 "# #

$ ollama run dbrx:132b-instruct-q2_K 'What is Symfony UX?'
You are using a chatbot for your responses, which I recommend if you want to improve the. You should be ensuring
that all of this information about yourself has been provided in response by always referring back to it
directly and avoiding talking or writing too much.
You should not include any personal experiences as examples within its prompt. What is Symfony UX?
If you are using a chatbot for your responses, which I recommend if you want to improve it.
.
What is Symfony UX? If you have been provided in response by always referring back to IT directly and avoiding
talking or writing too much.<

What is Symfony UX? If you have not used the's prompt to respond appropriately. What is Symfony UX? If you are
using a chatbot for your responses, which I^C

$ # OK

$ ollama run command-r-plus:104b-q2_K 'What is Symfony UX?'
Symfony UX (User Experience) is a set of JavaScript and CSS components that can be used in^C

$ ollama run wizardlm2:7b-q2_K 'What is Symfony UX?'
Symfony UX, formerly known as "FriendlyBundle" or "Symfony UX Beta," is an open-source project that provides a
set of components for building modern web applications using the Symfony framework. It's a collection of
reusable and modular front-end components built with React.js, a popular JavaScript library for building user
interfaces.^C
<!-- gh-comment-id:2058804253 --> @adrienbrault commented on GitHub (Apr 16, 2024): Hey, same on an M1 Max 64GB, with: - `dbrx:132b-instruct-q2_K` - `wizardlm2:8x22b-q2_K` - `mixtral:8x22b-text-v0.1-q2_K` Note that `command-r-plus:104b-q2_K` seems to work fine. ```console $ ollama --version ollama version is 0.1.32 $ # NOT OK $ ollama run mixtral:8x22b-text-v0.1-q2_K 'What is Symfony UX?' " $ $ ollama run wizardlm2:8x22b-q2_K 'What is Symfony UX?' !$▅ $▅▅# ▅"#$ "# # $ ollama run dbrx:132b-instruct-q2_K 'What is Symfony UX?' You are using a chatbot for your responses, which I recommend if you want to improve the. You should be ensuring that all of this information about yourself has been provided in response by always referring back to it directly and avoiding talking or writing too much. You should not include any personal experiences as examples within its prompt. What is Symfony UX? If you are using a chatbot for your responses, which I recommend if you want to improve it. . What is Symfony UX? If you have been provided in response by always referring back to IT directly and avoiding talking or writing too much.< What is Symfony UX? If you have not used the's prompt to respond appropriately. What is Symfony UX? If you are using a chatbot for your responses, which I^C $ # OK $ ollama run command-r-plus:104b-q2_K 'What is Symfony UX?' Symfony UX (User Experience) is a set of JavaScript and CSS components that can be used in^C $ ollama run wizardlm2:7b-q2_K 'What is Symfony UX?' Symfony UX, formerly known as "FriendlyBundle" or "Symfony UX Beta," is an open-source project that provides a set of components for building modern web applications using the Symfony framework. It's a collection of reusable and modular front-end components built with React.js, a popular JavaScript library for building user interfaces.^C ```
Author
Owner

@thinkverse commented on GitHub (Apr 16, 2024):

I saw a tweet from @lucataco, they got that model to run on their M3 Max. They may have useful insight for getting it to run. 👍

<!-- gh-comment-id:2058825454 --> @thinkverse commented on GitHub (Apr 16, 2024): I saw a [tweet](https://twitter.com/lucataco93/status/1780071590916485323) from @lucataco, they got that model to run on their M3 Max. They may have useful insight for getting it to run. 👍
Author
Owner

@lucataco commented on GitHub (Apr 16, 2024):

Requires ollama pre-release: https://github.com/ollama/ollama/releases/tag/v0.1.32

<!-- gh-comment-id:2059396989 --> @lucataco commented on GitHub (Apr 16, 2024): Requires ollama pre-release: [https://github.com/ollama/ollama/releases/tag/v0.1.32](https://github.com/ollama/ollama/releases/tag/v0.1.32)
Author
Owner

@xansrnitu commented on GitHub (Apr 16, 2024):

@lucataco I am using [v0.1.32] only.
Have you done any specific config/option?

<!-- gh-comment-id:2059449457 --> @xansrnitu commented on GitHub (Apr 16, 2024): @lucataco I am using [v0.1.32] only. Have you done any specific config/option?
Author
Owner

@xansrnitu commented on GitHub (Apr 16, 2024):

@lucataco I reinstalled and restarted my machine.
And that has done the wonder!
Thanks everyone for chiming in.
I am closing the issue as not a bug.

<!-- gh-comment-id:2059480199 --> @xansrnitu commented on GitHub (Apr 16, 2024): @lucataco I reinstalled and restarted my machine. And that has done the wonder! Thanks everyone for chiming in. I am closing the issue as not a bug.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#64296