Update app.py

app.py

@@ -1,4 +1,3 @@
-
 import os
 from flask import Flask, request, jsonify
 from llama_cpp import Llama
@@ -6,24 +5,42 @@ from huggingface_hub import hf_hub_download
 
 app = Flask(__name__)
 
-# Replace with your info
+# Model details
 REPO_ID = "CyberCoder225/maira-model" 
 FILENAME = "SmolLM2-360M-Instruct.Q4_K_M.gguf"
 
-# This downloads the model from HF to Render's temporary memory
+# 1. Downloads the model to the Space's storage
 print("Fetching Maira's brain from Hugging Face...")
 model_path = hf_hub_download(repo_id=REPO_ID, filename=FILENAME)
 
+# 2. Loads the model into RAM
 llm = Llama(model_path=model_path, n_ctx=2048)
 
+# --- ADDED THIS TO FIX THE 404 ---
+@app.route('/', methods=['GET'])
+def home():
+    return jsonify({
+        "status": "online",
+        "message": "Maira API is running! Send POST requests to /chat",
+        "model": FILENAME
+    })
+
 @app.route('/chat', methods=['POST'])
 def chat():
-    data = request.json
-    user_input = data.get("message", "")
-    prompt = f"### User: {user_input}\n### Maira:"
-    output = llm(prompt, max_tokens=150, stop=["###", "</s>"], echo=False)
-    response = output["choices"][0]["text"].strip()
-    return jsonify({"maira": response})
+    try:
+        data = request.json
+        user_input = data.get("message", "")
+        
+        # Simple prompt format for SmolLM
+        prompt = f"### User: {user_input}\n### Maira:"
+        
+        output = llm(prompt, max_tokens=150, stop=["###", "</s>"], echo=False)
+        response = output["choices"][0]["text"].strip()
+        
+        return jsonify({"maira": response})
+    except Exception as e:
+        return jsonify({"error": str(e)}), 500
 
 if __name__ == "__main__":
-    app.run(host="0.0.0.0", port=7860)
+    # --- FIXED PORT FOR HUGGING FACE ---
+    app.run(host="0.0.0.0", port=7860)