Update "Contribute" guide with uv and DevContainer intructions

[yvan-sraka]

Rendered

• Add instructions about different outlines "flavors"

[yvan-sraka]

@cpfiffer I'm not sure what's the best guide to update, but it could be nice to also add something like:

Optional Dependencies

Outlines provides multiple optional dependency sets to support different backends and use cases. You can install them as needed using:

• pip install ".[vllm]" for vLLM, optimized for high-throughput inference.
• pip install ".[transformers]" for Hugging Face Transformers.
• pip install ".[mlxlm]" for MLX-LM, optimized for Apple silicon.
• pip install ".[openai]" to use OpenAI’s API.
• pip install ".[llamacpp]" for llama.cpp, a lightweight LLM inference engine.
• pip install ".[exllamav2]" for ExLlamaV2, optimized for NVIDIA GPUs.

WDYT?

[cpfiffer]

I would stick the extra deps stuff in the installation guide but use pip install "outlines[vllm]" instead. Let's also stick an example that uses the dot-install method for one of the backends in the contribution guide, but point to available installation backends in the installaction guide.

[yvan-sraka]

I would stick the extra deps stuff in the installation guide but use pip install "outlines[vllm]" instead. Let's also stick an example that uses the dot-install method for one of the backends in the contribution guide, but point to available installation backends in the installaction guide.

@cpfiffer I'm not 100% sure if I can drop the first paragraph here then, that sounds a bit redundant, right?

[cpfiffer]

I'm not sure what it's redundant with. As far as I am aware, the paragraph you suggested is the only comprehensive list of supportive optional dependency commands.

Optional Dependencies

Outlines provides multiple optional dependency sets to support different backends and use cases. You can install them as needed using:

• pip install ".[vllm]" for vLLM, optimized for high-throughput inference.
• pip install ".[transformers]" for Hugging Face Transformers.
• pip install ".[mlxlm]" for MLX-LM, optimized for Apple silicon.
• pip install ".[openai]" to use OpenAI’s API.
• pip install ".[llamacpp]" for llama.cpp, a lightweight LLM inference engine.
• pip install ".[exllamav2]" for ExLlamaV2, optimized for NVIDIA GPUs.

Then, in the contribution guide, we would only need to refer to this list and highlight the single most common optional dependency to indicate what the users should expect to find in the installation guide.

[rlouf]

@yvan-sraka is it ready to merge?