Update app.py

app.py

@@ -1,37 +1,24 @@
 import gradio as gr
 from llama_cpp import Llama
-from huggingface_hub import hf_hub_download
+import os
 
-# 1. Download GGUF file from your Repo
-model_path = hf_hub_download(
-    repo_id="SALEETAI/coding-agent-qwen-sft-v3-GGUF",
-    filename="coding-agent-qwen-sft-v3-GGUF.q4_k_m.gguf" 
-)
+# Path is now local because we downloaded it in Dockerfile
+model_path = "/app/coding-agent-qwen-sft-v3-GGUF.q4_k_m.gguf"
 
-# 2. Initialize the model
-# We set n_ctx to 2048 for good performance on free CPU
-llm = Llama(model_path=model_path, n_ctx=2048, verbose=False)
+print(f"Checking for model at {model_path}...")
+if not os.path.exists(model_path):
+    print("MODEL NOT FOUND!")
 
-def generate_code(prompt):
-    # Professional inference parameters
+llm = Llama(model_path=model_path, n_ctx=2048)
+
+def generate(prompt):
     output = llm(
         f"<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant\n",
         max_tokens=1024,
         stop=["<|im_end|>"],
-        echo=False,
         repeat_penalty=1.2,
         temperature=0.4
     )
     return output['choices'][0]['text']
 
-# 3. Create Gradio UI
-demo = gr.Interface(
-    fn=generate_code,
-    inputs=gr.Textbox(lines=5, placeholder="Enter a coding prompt..."),
-    outputs=gr.Code(language="python"),
-    title="🚀 Coding Agent Qwen-SFT v2 (GGUF Demo)",
-    description="Running on Free CPU using llama-cpp-python."
-)
-
-if __name__ == "__main__":
-    demo.launch(server_name="0.0.0.0", server_port=7860)
+gr.ChatInterface(generate).launch(server_name="0.0.0.0", server_port=7860)