egutierrez
aef8791151
- Added Appshell component with responsive navbar and main content area - Integrated ColorSchemeToggle for light/dark mode switching - Created Welcome component with styled title and introductory text - Developed ChatPage for LLM interaction with WebSocket support - Implemented Biblioteca for managing notes with rich text editor - Added LoginPage for user authentication with error handling - Introduced MessageList and MessageBubble components for chat messages - Styled components with CSS modules for consistent design
68 lines
2.2 KiB
Python
68 lines
2.2 KiB
Python
from domains.Llms.Modelos.Base_model import ModeloABC
|
|
from domains.Security.GenerarIDs import GeneradorIDUnico
|
|
from typing import AsyncGenerator, Union
|
|
from domains.ConexionApis.Ollama_cliente import OllamaCliente # Asegúrate de importar correctamente
|
|
import asyncio
|
|
|
|
class ModeloOllama(ModeloABC):
|
|
def __init__(
|
|
self,
|
|
cliente: OllamaCliente,
|
|
model: str = "llama3",
|
|
id: str = None,
|
|
temperature: float = 0.7,
|
|
top_p: float = 1.0,
|
|
top_k: int = None,
|
|
frecuencia_penalizacion: float = 0.0,
|
|
num_tokens_maximos: int = 512
|
|
):
|
|
if not isinstance(cliente, OllamaCliente):
|
|
raise TypeError("El parámetro 'cliente' debe ser una instancia de OllamaCliente")
|
|
|
|
|
|
self.id = id if id else GeneradorIDUnico("MOOL").generar()
|
|
super().__init__(
|
|
model=model,
|
|
temperature=temperature,
|
|
top_p=top_p,
|
|
top_k=top_k,
|
|
frecuencia_penalizacion=frecuencia_penalizacion,
|
|
num_tokens_maximos=num_tokens_maximos
|
|
)
|
|
self.cliente = cliente
|
|
|
|
async def responder(
|
|
self,
|
|
prompt: str,
|
|
system_prompt: str = "",
|
|
stream: bool = False,
|
|
**kwargs
|
|
) -> Union[str, AsyncGenerator[str, None]]:
|
|
messages = []
|
|
if system_prompt:
|
|
messages.append({"role": "system", "content": system_prompt})
|
|
messages.append({"role": "user", "content": prompt})
|
|
|
|
def sync_call():
|
|
return self.cliente.chat_completion(
|
|
model=self.model,
|
|
messages=messages,
|
|
temperature=self.temperature,
|
|
top_p=self.top_p,
|
|
max_tokens=self.num_tokens_maximos,
|
|
frequency_penalty=self.frecuencia_penalizacion,
|
|
stream=stream,
|
|
**kwargs
|
|
)
|
|
|
|
loop = asyncio.get_event_loop()
|
|
resultado = await loop.run_in_executor(None, sync_call)
|
|
|
|
if stream:
|
|
async def generador():
|
|
for token in resultado:
|
|
yield token
|
|
return generador()
|
|
else:
|
|
return resultado.choices[0].message.content
|