Add note on local inference alternative to HfApiModel

[HKanoje]

This PR adds a note in the docs explaining how users can run models locally using Ollama + LiteLLM as an alternative to HfApiModel() in case they hit Hugging Face API credit limits.

It includes a sample code snippet using:

LiteLLMModel with model_id="ollama_chat/qwen2.5:7b"

It closes huggingface/smolagents#967 Issues

[davidberenstein1957]

Hi @HKanoje, this could be a nice addition if we could scope this a bit more uniform.

Could you create a uniform section on this within the onboarding? I think we can just mention local serving and ideally use similar API classed (the HF ones) but explain how to redirect the base URL that relies on OpenAIAPI spec for something like ollama.

WDYT?

[HKanoje]

@davidberenstein1957 Yes I think you're right. It makes more sense to put in as a different section because HfApiModel() is been used everywhere what do you suggest where should I make this section? During the onboarding steps?

[HKanoje]

Hey, @davidberenstein1957 I’ve added the local inference section to the onboarding as you suggested. Let me know if there’s anything you'd like me to adjust or improve! Check here #327