docs: Add LLM messageHistory

[jakmro]

Description

Add LLM messageHistory documentation

Type of change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update (improves or adds clarity to existing documentation)

Checklist

• I have performed a self-review of my code
• I have commented my code, particularly in hard-to-understand areas
• I have updated the documentation accordingly
• My changes generate no new warnings

[chmjkb]

Can we maybe also add here that it can be used to provide context to the model? I feel like this can be misunderstood

[kuriel-dev]

Hi everyone, first of all, thanks so much for this amazing tool.

Second, I’m here to report a bug related to the messageHistory.
(I’ve made sure this is set as stable as possible before reporting.)

I’ve thoroughly tested it, and the messageHistory appears to be stable because:

1. I’m loading a chat from a phone database.
2. I’m passing the messageHistory (mapped) to useLLM().
3. This only happens once. (i pass messageHistory only 1 time)

When messageHistory is empty, the AI works perfectly.

But when messageHistory has elements, the AI gone crazy.

I’ve tried removing messageHistory and it works fine, but without context, it can’t resume the conversation.

const llama = useLLM({
  modelSource: modelSource,
  tokenizerSource: tokenizerSource,
  // messageHistory: messageHistory,
});

an example of messageHistory i'm passing:

    [
      {"content": "Hola cómo estás?", "role": "user"}, 
      {"content": "Estoy bien, gracias. ¿En qué puedo ayudarte?", "role": "assistant"}
    ]

Do you have any advice on how to properly resume the conversation?

[kuriel-dev]

I also think that using useLLM should have deps array, to make it stable:

const llama = useLLM({
    modelSource:  // the modelSource,
    tokenizerSource: // the tokenizerSource,
  }, [deps]); <---- array deps for best sync

Example:

const llama = useLLM({
    modelSource,
    tokenizerSource,
   messageHistory,
  }, [messageHistory]);

since llama.setMessageHistory does not exists. (this can be another good option)