Add support for Llama2, Palm, Cohere, Anthropic, Replicate, Azure Models - using litellm

[ishaan-jaff]

This PR adds support for 50+ models with a standard I/O interface using: https://github.com/BerriAI/litellm/

ChatLiteLLM() is integrated into langchain and allows you to call all models using the ChatOpenAI I/O interface
https://python.langchain.com/docs/integrations/chat/litellm

Here's an example of how to use ChatLiteLLM()

ChatLiteLLM(model="gpt-3.5-turbo")
ChatLiteLLM(model="claude-2", temperature=0.3)
ChatLiteLLM(model="command-nightly")
ChatLiteLLM(model="replicate/llama-2-70b-chat:2c1608e18606fad2812020dc541930f2d0495ce32eee50074220b87300bc16e1")

[ishaan-jaff]

Addressing: #9

[ishaan-jaff]

@fynnfluegge can you please take a look at this PR when possible?

Happy to add docs/testing if this initial commit looks good to you

If this PR looks useful would love to meet to discuss how we can be helpful to you. Here's my calendly for your convenience

[fynnfluegge]

@ishaan-jaff this is great, thanks a lot! 🔥
I think a new argarse argument would make sense here, instead the --gpt4 argument. Maybe --model with default gpt3.5-turbo as currently. Then any litellm supported model can be used with that argument. Maybe a section in the README.md would be helpful.

One question: litellm uses only environment parameters to access sagemaker or azure hosted models right?
I also prepared a small PR that exposes the sagemaker and azure langchain api here #12 but I guess this one is absolete if we use litellm, isn't it?

[fynnfluegge]

@ishaan-jaff tested with OpenAI, works great! If you could run poetry lock in your branch I will merge this 🙌