[GH-ISSUE #1499] Add mistral's new 7B-instruct-v0.2 #62847

Closed
opened 2026-05-03 10:29:21 -05:00 by GiteaMirror · 4 comments
Owner

Originally created by @tarek-ayed on GitHub (Dec 13, 2023).
Original GitHub issue: https://github.com/ollama/ollama/issues/1499

Along with many releases, Mistral vastly improved their existing 7B model with a version named v0.2.

It has 32k context instead of 8k and better benchmark scores: https://x.com/dchaplot/status/1734198245067243629?s=20
More can be found here: https://docs.mistral.ai/platform/endpoints (see "Mistral Tiny")

The weights are published on HuggingFace:

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

I don't know if anything needs to be implemented on the llama.cpp front, given that it's the same architecture as before.

Let me know how I can contribute to make this happen ;)

Originally created by @tarek-ayed on GitHub (Dec 13, 2023). Original GitHub issue: https://github.com/ollama/ollama/issues/1499 Along with many releases, Mistral vastly improved their existing 7B model with a version named `v0.2`. It has 32k context instead of 8k and better benchmark scores: https://x.com/dchaplot/status/1734198245067243629?s=20 More can be found here: https://docs.mistral.ai/platform/endpoints (see "Mistral Tiny") The weights are published on HuggingFace: https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2 I don't know if anything needs to be implemented on the llama.cpp front, given that it's the same architecture as before. Let me know how I can contribute to make this happen ;)
Author
Owner

@justinh-rahb commented on GitHub (Dec 13, 2023):

No changes needed, was already uploaded yesterday, just repull Mistral

ollama pull mistral
<!-- gh-comment-id:1853839957 --> @justinh-rahb commented on GitHub (Dec 13, 2023): No changes needed, was already uploaded yesterday, just repull Mistral ``` ollama pull mistral ```
Author
Owner

@mchiang0610 commented on GitHub (Dec 13, 2023):

Closing since this is available now. Sorry about that. It means that we'll have to really improve the Ollama library to make those changes much more visible.

<!-- gh-comment-id:1854244147 --> @mchiang0610 commented on GitHub (Dec 13, 2023): Closing since this is available now. Sorry about that. It means that we'll have to really improve the Ollama library to make those changes much more visible.
Author
Owner

@Liuxyly commented on GitHub (Dec 14, 2023):

What about this version on HuggingFace?
https://huggingface.co/mistralai/Mixtral-8x7B-v0.1

<!-- gh-comment-id:1855014836 --> @Liuxyly commented on GitHub (Dec 14, 2023): What about this version on HuggingFace? https://huggingface.co/mistralai/Mixtral-8x7B-v0.1
Author
Owner

@justinh-rahb commented on GitHub (Dec 14, 2023):

@Liuxyly The PR #1475 for that model's support is merged to main but a new release hasn't been created yet. I'm sure it will be soon, and the model image to go with it.

<!-- gh-comment-id:1855053306 --> @justinh-rahb commented on GitHub (Dec 14, 2023): @Liuxyly The PR #1475 for that model's support is merged to main but a new release hasn't been created yet. I'm sure it will be soon, and the model image to go with it.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#62847