Update app.py

app.py

@@ -13,7 +13,7 @@ from fastapi import FastAPI, HTTPException
 from fastapi.responses import StreamingResponse
 from fastapi.middleware.cors import CORSMiddleware
 from pydantic import BaseModel
-from typing import List, Optional, AsyncGenerator
+from typing import List, Optional, AsyncGenerator  # ✅ AsyncGenerator import
 import asyncio
 import gradio as gr
 
@@ -361,7 +361,7 @@ print(f"\n🎉 Ready! Default model: {current_backend.get_name()}")
 # Streaming Generator
 # ==============================================================================
 
-async def generate_stream(prompt: str, backend, temperature: float) -> AsyncGenerator[str]:
+async def generate_stream(prompt: str, backend, temperature: float) -> AsyncGenerator[str, None]:  # ✅ Fixed type hint
     encoded_prompt = tokenizer.encode(prompt)
     input_ids = [i for i in encoded_prompt.ids if i != eos_token_id]
     generated = input_ids.copy()
@@ -460,7 +460,7 @@ async def chat_completions(request: ChatCompletionRequest):
                 }]
             }
             yield f"data: {json.dumps(chunk)}\n\n"
-        yield "data: [DONE]\n\n"
+        yield " [DONE]\n\n"
 
     if request.stream:
         return StreamingResponse(event_stream(), media_type="text/event-stream")
@@ -468,7 +468,7 @@ async def chat_completions(request: ChatCompletionRequest):
         full = ""
         async for token in event_stream():
             if "[DONE]" not in token:
-                data = json.loads(token.replace("data: ", "").strip())
+                data = json.loads(token.replace(" ", "").strip())
                 full += data["choices"][0]["delta"]["content"]
         return {"choices": [{"message": {"content": full}}]}