change method

app.py

@@ -1,17 +1,53 @@
-import gradio as gr
-from transformers import AutoModelForCausalLM, AutoTokenizer
+from fastapi import FastAPI, HTTPException
+from pydantic import BaseModel
+from huggingface_hub import InferenceClient
+from typing import List, Tuple
 
-# Charger un modèle Hugging Face (ici GPT-J)
-model_name = "HuggingFaceH4/zephyr-7b-beta"
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForCausalLM.from_pretrained(model_name)
+# Initialisation du client Hugging Face
+client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
 
-# Fonction pour générer du texte
-def generate_text(prompt):
-    inputs = tokenizer(prompt, return_tensors="pt")
-    outputs = model.generate(inputs["input_ids"], max_length=100, num_return_sequences=1)
-    return tokenizer.decode(outputs[0], skip_special_tokens=True)
+# Initialisation de FastAPI
+app = FastAPI()
 
-# Interface Gradio
-iface = gr.Interface(fn=generate_text, inputs="text", outputs="text")
-iface.launch()
+# Modèle pour les données d'entrée
+class PredictionRequest(BaseModel):
+    message: str
+    history: List[Tuple[str, str]] = []
+    system_message: str = "You are a friendly Chatbot."
+    max_tokens: int = 512
+    temperature: float = 0.7
+    top_p: float = 0.95
+
+@app.post("/predict")
+async def predict(request: PredictionRequest):
+    """
+    Endpoint REST pour effectuer une prédiction.
+    """
+    # Préparer les messages pour l'inférence
+    messages = [{"role": "system", "content": request.system_message}]
+    for user_input, assistant_response in request.history:
+        if user_input:
+            messages.append({"role": "user", "content": user_input})
+        if assistant_response:
+            messages.append({"role": "assistant", "content": assistant_response})
+    messages.append({"role": "user", "content": request.message})
+    
+    # Appel de l'API Hugging Face
+    try:
+        response = ""
+        for message in client.chat_completion(
+            messages,
+            max_tokens=request.max_tokens,
+            stream=True,
+            temperature=request.temperature,
+            top_p=request.top_p,
+        ):
+            response += message.choices[0].delta.content
+        return {"response": response}
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))
+
+# Pour le test en local
+if __name__ == "__main__":
+    import uvicorn
+    uvicorn.run(app, host="0.0.0.0", port=7860)

requirements.txt

@@ -1,4 +1,3 @@
-huggingface_hub==0.25.2
-transformers
-gradio
-torch==1.13.1
+fastapi
+uvicorn
+huggingface_hub