Update app.py

app.py

@@ -1,54 +1,75 @@
 import gradio as gr
-from threading import Thread
-from transformers import AutoTokenizer, AutoModelForCausalLM, TextIteratorStreamer
-import torch
+from transformers import pipeline
+import os
+import shutil
 
-# Load non-gated model and tokenizer
-model_id = "Qwen/Qwen2-7B-Instruct"
-tokenizer = AutoTokenizer.from_pretrained(model_id)
-model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.bfloat16, device_map="auto")
+# Set HF_HOME to persistent storage to avoid temporary storage bloat
+os.environ["HF_HOME"] = "/data/.huggingface"
 
-def chat(message, history):
-    # Build message history
-    messages = [
-        {"role": "system", "content": "You are a helpful and friendly assistant."}
-    ] + history + [{"role": "user", "content": message}]
-    
-    # Prepare inputs
-    inputs = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt").to(model.device)
-    
-    # Set up streamer for live typing
-    streamer = TextIteratorStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
-    
-    # Generation kwargs
-    generation_kwargs = {
-        "inputs": inputs,
-        "streamer": streamer,
-        "max_new_tokens": 256,
-        "do_sample": True,
-        "top_p": 0.95,
-        "temperature": 0.7,
-    }
+# Load open-source model for code generation (no token required)
+code_generator = pipeline("text-generation", model="bigcode/starcoder")
+
+def generate_code_and_response(prompt):
+    # Generate code using the model
+    code_prompt = f"Generate Python code for: {prompt}\n```python\n"
+    code_output = code_generator(code_prompt, max_length=300, num_return_sequences=1, do_sample=True, temperature=0.7)[0]['generated_text']
     
-    # Run generation in a separate thread
-    thread = Thread(target=model.generate, kwargs=generation_kwargs)
-    thread.start()
+    # Extract the code part (between ```python and ```)
+    code_start = code_output.find("```python") + 9
+    code_end = code_output.find("```", code_start)
+    code = code_output[code_start:code_end].strip() if code_end != -1 else code_output.strip()
+
+    # Generate conversational response in Grok-like style
+    response_text = (
+        f"Alright, I’ve spun up some Python code for your request: '{prompt}'. Here it comes:\n\n"
+        f"```python\n{code}\n```\n\n"
+        f"Pretty cool, right? I’m running on an open-source model, zipping along on Hugging Face’s free tier. "
+        f"Want me to tweak this code, explain it, or whip up something else? I’m all cosmic curiosity!"
+    )
     
-    # Yield tokens for live streaming
-    generated_text = ""
-    for new_text in streamer:
-        generated_text += new_text
-        yield generated_text
+    # Clean up cache to avoid storage limit issues
+    cache_dir = os.path.join("/data/.huggingface", "hub")
+    if os.path.exists(cache_dir):
+        shutil.rmtree(cache_dir, ignore_errors=True)
+
+    # Escape special characters for JavaScript
+    escaped_response = response_text.replace("`", "\\`").replace("\n", "\\n")
     
-    thread.join()
+    # Wrap response in HTML with JavaScript for typing effect
+    html_response = """
+    <div id="typing-output" style="font-family: Arial, sans-serif; font-size: 16px; white-space: pre-wrap;"></div>
+    <script>
+        const text = '""" + escaped_response + """';
+        let index = 0;
+        const speed = 30; // Typing speed in milliseconds
+        const outputElement = document.getElementById('typing-output');
+        
+        function typeWriter() {
+            if (index < text.length) {
+                outputElement.innerHTML += text.charAt(index);
+                index++;
+                setTimeout(typeWriter, speed);
+            }
+        }
+        typeWriter();
+    </script>
+    """
+    return html_response
 
-# Create Gradio chat interface
-demo = gr.ChatInterface(
-    fn=chat,
-    type="messages",
-    title="Qwen2-7B Chatbot",
-    description="Chat with a non-gated Qwen2-7B-Instruct model. Responses stream live.",
-    examples=[["Tell me a fun fact."], ["Explain neural networks in simple terms."]],
+# Create Gradio interface
+demo = gr.Interface(
+    fn=generate_code_and_response,
+    inputs=gr.Textbox(label="Enter your coding request", placeholder="e.g., Create a function to calculate factorial"),
+    outputs=gr.HTML(label="Generated Code and Response"),
+    title="Live CodeGen Bot",
+    description="Generate Python code using open-source models with a Grok-like conversational style and live typing effect. Hosted on Hugging Face Spaces free tier.",
+    css="""
+        body { background-color: #f0f2f5; }
+        .gr-box { border-radius: 8px; padding: 20px; }
+        #typing-output { background: #fff; padding: 15px; border: 1px solid #ddd; border-radius: 5px; }
+    """
 )
 
-demo.launch()
+# Launch the app
+if __name__ == "__main__":
+    demo.launch()