Update app.py

app.py

@@ -5,7 +5,7 @@ from fastapi.staticfiles import StaticFiles
 from pathlib import Path
 from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
 
-# Load a decoder-only model like Falcon
+# Model: Falcon-rw-1b (decoder-only, not instruction-tuned)
 model_id = "tiiuae/falcon-rw-1b"
 tokenizer = AutoTokenizer.from_pretrained(model_id)
 model = AutoModelForCausalLM.from_pretrained(model_id)
@@ -15,15 +15,15 @@ pipe = pipeline("text-generation", model=model, tokenizer=tokenizer)
 # Init FastAPI app
 app = FastAPI()
 
-# Mount static directory (serves script.js, index.html)
+# Mount static directory
 app.mount("/static", StaticFiles(directory="."), name="static")
 
-# Serve the frontend page
+# Serve HTML UI
 @app.get("/", response_class=HTMLResponse)
 async def serve_page():
     return HTMLResponse(Path("index.html").read_text())
 
-# Chatbot API route
+# Chat API
 @app.post("/api")
 async def ask_ai(request: Request):
     data = await request.json()
@@ -33,9 +33,8 @@ async def ask_ai(request: Request):
         return JSONResponse(content={"answer": "❗ Please enter a valid question."})
 
     try:
-        # Decoder-only models generate continuations; use prompt format
-        prompt = f"[INST] {question} [/INST]"
-        output = pipe(prompt, max_new_tokens=256)[0]["generated_text"]
+        # Falcon works with plain prompts
+        output = pipe(question, max_new_tokens=256, return_full_text=False)[0]["generated_text"]
         return JSONResponse(content={"answer": output.strip()})
     except Exception as e:
         return JSONResponse(content={"answer": f"⚠️ Error: {str(e)}"})