agents/tools/llm_tools.py

#from .base_tool import BaseTool
from models.tinygpt2_model import TinyGPT2Model
from pydantic import PrivateAttr
from crewai.tools import BaseTool



class MistralChatTool(BaseTool):
    name: str = "mistral_chat"
    description: str = "Generate an empathetic AI chat response."
    model_config = {"arbitrary_types_allowed": True}
    _model: TinyGPT2Model = PrivateAttr()
    def __init__(self, config=None):
        super().__init__()
        self._model = TinyGPT2Model()
    def _run(self, prompt: str, context: dict = None):
        msg = f"Context: {context}\nUser: {prompt}" if context else prompt
        return self.model.generate(msg)

class GenerateAdviceTool(BaseTool):
    name: str = "generate_advice"
    description: str = "Generate personalized advice."
    model_config = {"arbitrary_types_allowed": True}
    _model: TinyGPT2Model = PrivateAttr()
    def __init__(self, config=None):
        super().__init__()
        self._model = TinyGPT2Model()
    def _run(self, user_analysis: dict, wisdom_quotes: list):
        prompt = f"Advice for: {user_analysis}, with wisdom: {wisdom_quotes}"
        return self.model.generate(prompt, max_length=300)

class SummarizeConversationTool(BaseTool):
    name: str = "summarize_conversation"
    description: str = "Summarize chat with insights and next steps."
    model_config = {"arbitrary_types_allowed": True}
    _model: TinyGPT2Model = PrivateAttr()
    def __init__(self, config=None):
        super().__init__()
        self._model = TinyGPT2Model()
    def _run(self, conversation: list):
        prompt = f"Summarize: {conversation}"
        return self.model.generate(prompt, max_length=200)

class LLMTools:
    def __init__(self, config=None):
        self.mistral_chat = MistralChatTool(config)
        self.generate_advice = GenerateAdviceTool(config)
        self.summarize_conversation = SummarizeConversationTool(config)