Integrate mistral.rs LLM

[EricLBuehler]

Description

In this PR, I have added support for the mistral.rs LLM inference platform via a new integration. mistral.rs is a new LLM inference platform with key features such as prefix caching, optimized X-LoRA support, LoRA support via weight merging and grammar support.

New Package?

Did I fill in the tool.llamahub section in the pyproject.toml and provide a detailed README.md for my new integration or package?

• [x] Yes
• [ ] No

Version Bump?

Did I bump the version in the pyproject.toml file of the package I am updating? (Except for the llama-index-core package)

• [ ] Yes
• [x] No

Type of Change

Please delete options that are not relevant.

• [x] New feature (non-breaking change which adds functionality)

How Has This Been Tested?

Please describe the tests that you ran to verify your changes. Provide instructions so we can reproduce. Please also list any relevant details for your test configuration

• [ ] Added new unit/integration tests
• [x] Added new notebook (that tests end-to-end)
• [x] I stared at the code and made sure it makes sense

Suggested Checklist:

• [x] I have performed a self-review of my own code
• [x] I have commented my code, particularly in hard-to-understand areas
• [x] I have made corresponding changes to the documentation
• [ ] I have added Google Colab support for the newly added notebooks.
• [x] My changes generate no new warnings
• [ ] I have added tests that prove my fix is effective or that my feature works
• [ ] New and existing unit tests pass locally with my changes
• [x] I ran make format; make lint to appease the lint gods

[review-notebook-app[bot]]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[EricLBuehler]

Thank you! I've removed formatting the whole codebase and just formatted my new changes.

[nerdai]

Thank you! I've removed formatting the whole codebase and just formatted my new changes.

thanks!

[nerdai]

@EricLBuehler do you mind adding me to your fork? Looks like we need to do some pants related stuff (i.e. run pants tailor :: in the root of your project)

[EricLBuehler]

@nerdai, I addressed your comments and have added you to the repo.

[nerdai]

@EricLBuehler looks like we're running into error still:

E   SyntaxError: keyword argument repeated: logprobs

[EricLBuehler]

@nerdai, sorry for that mistake. It should be fixed now.

[nerdai]

@nerdai, sorry for that mistake. It should be fixed now.

All good -- thanks for the quick fix!

[EricLBuehler]

It seems like the CI tests are failing because this integration depends on the mistralrs library. Would it be best if I update the CI to install and build mistralrs?

[nerdai]

It seems like the CI tests are failing because this integration depends on the mistralrs library. Would it be best if I update the CI to install and build mistralrs?

yes, please ensure all the required deps are listed in the pyproject.toml. Best to run:

poetry add mistralrs

have you published mistralrs to pypi yet?

[EricLBuehler]

@nerdai, I just released mistralrs on pypi. However, it requires a Rust toolchain to build. Can we update the CI to install the Rust toolchain?

[nerdai]

@nerdai, I just released mistralrs on pypi. However, it requires a Rust toolchain to build. Can we update the CI to install the Rust toolchain?

Ah okay. Can it work with just he standard rust installation?

[EricLBuehler]

@nerdai, yes, it can. It depends on openssl though.

[nerdai]

@nerdai, yes, it can. It depends on openssl though.

Sorry @EricLBuehler not sure if I'm following. To my knowledge, Rust Toolchain is installed in our github runners by default (source).

Can we not just do a poetry add mistralrs so that it gets added as a dep in pyproject.toml?

[EricLBuehler]

@nerdai, thanks for clarifying! I have added it as a dependency now.

[EricLBuehler]

Hi @nerdai! I have updated this PR to use our latest PyPi release. Additionally, I made sure the tests pass by running the following commands:

make format
make lint
pants tailor --check ::
poetry run make -s test

I think that the CI tests should pass now.

[nerdai]

@EricLBuehler we're getting close! Looks like tests are still failing. From the traceback captured in the logs, i see this:

    ) -> list[dict[str, str]]:
E   TypeError: 'type' object is not subscriptable

Maybe we need to do the following:

• change list to List in all such occurrences in base.py
• change dict to Dict in all such occurrences in base.py

where List and Dict are both from typing module.

[EricLBuehler]

@nerdai, that should be fixed now! Not sure why the tests I ran locally didn't catch that though.

[nerdai]

@nerdai, that should be fixed now! Not sure why the tests I ran locally didn't catch that though.

Happens to me too sometimes. I think there's a mismatch between versions perhaps on python or other testing/formatting dependencies. 🤔

[nerdai]

@EricLBuehler time to 🛳️!

Thanks for this :)

[EricLBuehler]

@nerdai thank you!