Update app.py

app.py

@@ -1,32 +1,28 @@
-import gradio as gr
+from fastapi import FastAPI
+from pydantic import BaseModel
 from huggingface_hub import InferenceClient
 
-# Modelo gratuito y público
+app = FastAPI()
 MODEL_ID = "google/gemma-2b-it"
 client = InferenceClient(model=MODEL_ID)
 
-def chat(message, history):
-    history = history or []
-    # Convierte el historial a formato texto
+class ChatRequest(BaseModel):
+    message: str
+    history: list[list[str]] | None = None
+
+@app.post("/chat")
+def chat_endpoint(request: ChatRequest):
+    history = request.history or []
     prompt = ""
     for user_msg, bot_msg in history:
         prompt += f"User: {user_msg}\nAssistant: {bot_msg}\n"
-    prompt += f"User: {message}\nAssistant:"
+    prompt += f"User: {request.message}\nAssistant:"
 
     try:
         response = ""
         for chunk in client.text_generation(prompt, max_new_tokens=128, stream=True):
             response += chunk.token.text
-        history.append((message, response.strip()))
-        return history, history
+        history.append((request.message, response.strip()))
+        return {"response": response.strip(), "history": history}
     except Exception as e:
-        return history + [(message, f"⚠️ Error interno: {e}")], history
-
-chatbot = gr.ChatInterface(
-    fn=chat,
-    title="Chat FastAPI con HF",
-    description="Chat simple usando google/gemma-2b-it en Hugging Face"
-)
-
-if __name__ == "__main__":
-    chatbot.launch()
+        return {"response": f"⚠️ Error interno: {e}", "history": history}