Create app.py

app.py

@@ -0,0 +1,69 @@
+import os
+import gradio as gr
+from huggingface_hub import hf_hub_download
+from llama_cpp import Llama
+
+# 🔧 CONFIGURATION: Change these to swap models
+MODEL_REPO = "bartowski/Qwen2.5-Coder-1.5B-Instruct-GGUF"
+MODEL_FILE = "Qwen2.5-Coder-1.5B-Instruct-Q4_K_M.gguf"
+# For 7B: MODEL_REPO = "bartowski/Qwen2.5-Coder-7B-Instruct-GGUF"
+#          MODEL_FILE = "Qwen2.5-Coder-7B-Instruct-Q4_K_M.gguf"
+
+# 1️⃣ Download model on first boot (cached automatically)
+print(f"⬇️ Downloading {MODEL_FILE} from {MODEL_REPO}...")
+model_path = hf_hub_download(repo_id=MODEL_REPO, filename=MODEL_FILE)
+
+# 2️⃣ Initialize CPU-optimized LLM
+llm = Llama(
+    model_path=model_path,
+    n_ctx=4096,          # Max context window
+    n_threads=2,         # Matches HF free tier vCPU count
+    n_batch=512,
+    verbose=False,
+    use_mlock=True       # Keep model in RAM (prevents swapping)
+)
+
+# 3️⃣ Generation function
+def generate_python_code(user_prompt):
+    system_prompt = (
+        "You are an expert Python developer. Write clean, efficient, PEP-8 compliant code. "
+        "Include type hints, docstrings, and error handling where appropriate. "
+        "Output only the code block unless explicitly asked for explanations."
+    )
+    
+    messages = [
+        {"role": "system", "content": system_prompt},
+        {"role": "user", "content": user_prompt}
+    ]
+    
+    output = llm.create_chat_completion(
+        messages=messages,
+        max_tokens=1024,
+        temperature=0.2,      # Low for deterministic code
+        top_p=0.9,
+        repeat_penalty=1.1,
+        stop=["</s>", "```"]  # Prevent runaway generation
+    )
+    
+    return output["choices"][0]["message"]["content"]
+
+# 4️⃣ Gradio UI
+demo = gr.Interface(
+    fn=generate_python_code,
+    inputs=gr.Textbox(
+        lines=4, 
+        placeholder="e.g., Write an async function to fetch JSON from a URL, retry 3 times on failure, and parse specific fields...",
+        label="Python Task"
+    ),
+    outputs=gr.Code(language="python", label="Generated Code"),
+    title="🐍 Python Dev Assistant",
+    description=f"Running `{MODEL_FILE}` on HF Free CPU Tier. First load takes ~60s.",
+    examples=[
+        ["Write a FastAPI route that accepts a CSV file and returns summary statistics"],
+        ["Refactor this list comprehension into a more readable loop with logging: `results = [x**2 for x in data if x > 0]`"],
+        ["Create a Pydantic model for a user profile with email validation and a custom validator for age > 18"]
+    ]
+)
+
+if __name__ == "__main__":
+    demo.launch()