s

app.py

@@ -1,23 +1,30 @@
-from fastapi import FastAPI, Request
-from fastapi.responses import JSONResponse
+from fastapi import FastAPI
+from fastapi.middleware.cors import CORSMiddleware
+from pydantic import BaseModel
 from model import generate_structure
 import uvicorn
 
 app = FastAPI()
 
-@app.get("/")
-def index():
-    return {"message": "Minecraft AI Builder Backend is running."}
+# Allow all CORS (for testing or frontend use)
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+class PromptRequest(BaseModel):
+    prompt: str
 
 @app.post("/prompt")
-async def prompt_handler(req: Request):
+async def prompt_route(data: PromptRequest):
     try:
-        data = await req.json()
-        prompt = data.get("prompt", "")
-        response = generate_structure(prompt)
+        response = generate_structure(data.prompt)
         return {"response": response}
     except Exception as e:
-        return JSONResponse(status_code=500, content={"error": str(e)})
+        return {"error": str(e)}
 
-if __name__ == "__main__":
-    uvicorn.run(app, host="0.0.0.0", port=7860)
+if __name__  == "__main__":
+    uvicorn.run(app, host="0.0.0.0", port=7860)

model.py

@@ -2,41 +2,38 @@ import os
 import requests
 from llama_cpp import Llama
 
-HF_TOKEN = os.getenv("HF_TOKEN")
-MODEL_REPO = "afrideva/TinyMistral-248M-SFT-v4-GGUF"
-MODEL_FILENAME = "tinymistral-248m-sft-v4.q2_k.gguf"
+MODEL_URL = "https://huggingface.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF/resolve/main/Llama-3.2-1B-Instruct.Q4_K_M.gguf"
+MODEL_FILENAME = "Llama-3.2-1B-Instruct.Q4_K_M.gguf"
 MODEL_PATH = f"./models/{MODEL_FILENAME}"
 
-# Manual download with fallback
-if not os.path.exists(MODEL_PATH):
-    print("📦 Downloading GGUF model manually from Hugging Face...")
-
-    url = f"https://huggingface.co/{MODEL_REPO}/resolve/main/{MODEL_FILENAME}"
-    headers = {"Authorization": f"Bearer {HF_TOKEN}"}
-
+def download_model():
     os.makedirs("./models", exist_ok=True)
-    with requests.get(url, headers=headers, stream=True) as r:
+    print("📦 Downloading GGUF model directly from URL...")
+    with requests.get(MODEL_URL, stream=True) as r:
         r.raise_for_status()
         with open(MODEL_PATH, "wb") as f:
             for chunk in r.iter_content(chunk_size=8192):
                 f.write(chunk)
-
     print(f"✅ Model downloaded to {MODEL_PATH}")
 
-# Load with llama-cpp
+# Only download if not already present
+if not os.path.exists(MODEL_PATH):
+    download_model()
+
+# Load model with llama-cpp-python
 llm = Llama(
     model_path=MODEL_PATH,
-    n_ctx=256,
-    n_threads=2,
-    n_batch=32,
-    n_gpu_layers=0,
-    chat_format=None  # Not using llama-2 format
+    n_ctx=512,
+    n_batch=512,
+    n_threads=6,
+    chat_format="llama-3",
+    verbose=False
 )
 
 def generate_structure(prompt: str) -> str:
-    output = llm.create_completion(
-        prompt=prompt,
-        temperature=0.7,
-        max_tokens=512,
-    )
-    return output["choices"][0]["text"].strip()
+    messages = [
+        {"role": "system", "content": "You are a Minecraft-style structure planner. You always respond with strictly valid JSON describing a 3D structure."},
+        {"role": "user", "content": prompt}
+    ]
+    output = llm.create_chat_completion(messages=messages)
+    return output["choices"][0]["message"]["content"]