Add MistralAILLM
#293
Add MistralAILLM
#293
Conversation
|
Docs are work in progress |
| output = [] | ||
| for response in responses: | ||
| chat_completion = response.choices[0] | ||
| try: |
There was a problem hiding this comment.
I believe this might be missing in some other implementations?
There was a problem hiding this comment.
will take a look, but maybe this is a special behaviour of this API, it doesn't deal with num_generations in the same way as OpenAI does, so it's easier to grab that choice
|
There's a version conflict with httpx between |
|
Related to argilla-io/argilla#4531 |
pyproject.toml
Outdated
| @@ -42,6 +42,7 @@ vertexai = ["google-cloud-aiplatform >= 1.38.0"] | |||
| together = ["together"] | |||
| argilla = ["argilla >= 1.22.0", "sentence-transformers >= 2.0.0", "textdescriptives >= 2.0.0"] | |||
There was a problem hiding this comment.
Shouldn't we pin Argilla to a higher version here?
There was a problem hiding this comment.
Is there a higher version than argilla>=1.22 right now?
There was a problem hiding this comment.
No, I mean when we merge the httpx bumped version
|
This should be ready once we have argilla-io/argilla#4531 merged |
…feat/mistral-api
Description
This PR integrates Mistral AI API as a new LLM.
The new
MistralAILLMis a wrapper over their python client.Closes #153