Create app.py

app.py

@@ -0,0 +1,47 @@
+import gradio as gr
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import torch
+
+model_id = "Qwen/Qwen2.5-Coder-32B-Instruct"
+
+# Load model and tokenizer
+tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
+model = AutoModelForCausalLM.from_pretrained(
+    model_id,
+    trust_remote_code=True,
+    device_map="auto",
+    torch_dtype=torch.float32  # CPU compatible
+)
+
+# Prompt formatter
+def generate_html_code(prompt):
+    instruction = f"Write only valid HTML code for this: {prompt}\n"
+    inputs = tokenizer(instruction, return_tensors="pt")
+    outputs = model.generate(
+        **inputs,
+        max_new_tokens=300,
+        temperature=0.7,
+        do_sample=True,
+        top_p=0.95
+    )
+    html_code = tokenizer.decode(outputs[0], skip_special_tokens=True)
+
+    # Optional: clean up output to extract HTML only
+    start = html_code.find("<")
+    if start != -1:
+        html_code = html_code[start:]
+    return html_code, html_code  # code + live preview
+
+# Gradio UI
+demo = gr.Interface(
+    fn=generate_html_code,
+    inputs=gr.Textbox(label="Describe your HTML component", placeholder="e.g. A red button that says Click Me", lines=3),
+    outputs=[
+        gr.Code(label="Generated HTML Code", language="html"),
+        gr.HTML(label="Live Preview")
+    ],
+    title="HTML Generator with Qwen2.5",
+    description="Generate and preview HTML in real-time using the Qwen2.5 32B Coder model (CPU mode, may be slow!)"
+)
+
+demo.launch()