Delete app.py.bak

app.py.bak

@@ -1,125 +0,0 @@
-import gradio as gr
-import torch
-from transformers import AutoTokenizer, AutoModelForCausalLM
-from peft import PeftModel
-import gc
-import uvicorn      # ← new
-
-# Model configuration
-MODEL_NAME = "robertnetwork/strudel-small"
-MAX_LENGTH = 2048
-
-# Global variables for model and tokenizer
-model = None
-tokenizer = None
-
-def load_model():
-    """Load the model and tokenizer"""
-    global model, tokenizer
-    
-    print("Loading model and tokenizer...")
-    
-    # Load tokenizer
-    tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
-    tokenizer.pad_token = tokenizer.eos_token
-    tokenizer.padding_side = "left"
-    
-    # Load model (CPU only for Spaces)
-    model = AutoModelForCausalLM.from_pretrained(
-        MODEL_NAME,
-        torch_dtype=torch.float32,
-        device_map="auto",
-        low_cpu_mem_usage=True
-    )
-    
-    print("Model loaded successfully!")
-
-def generate_code(instruction, style_input, max_new_tokens=512, temperature=0.7, top_p=0.9, do_sample=True):
-    """Generate Strudel.cc JavaScript code based on style input"""
-    if model is None or tokenizer is None:
-        return "Error: Model not loaded. Please wait for the model to initialize."
-    try:
-        prompt = f"Instruction: {instruction}\nInput: {style_input}\nOutput:"
-        inputs = tokenizer(
-            prompt,
-            return_tensors="pt",
-            truncation=True,
-            max_length=MAX_LENGTH - max_new_tokens,
-            padding=True
-        )
-        with torch.no_grad():
-            outputs = model.generate(
-                **inputs,
-                max_new_tokens=min(max_new_tokens, 256),
-                temperature=temperature,
-                top_p=top_p,
-                do_sample=do_sample,
-                pad_token_id=tokenizer.eos_token_id,
-                eos_token_id=tokenizer.eos_token_id,
-                repetition_penalty=1.1,
-                early_stopping=True,
-                no_repeat_ngram_size=3
-            )
-        full_response = tokenizer.decode(outputs[0], skip_special_tokens=True)
-        if "Output:" in full_response:
-            generated_code = full_response.split("Output:")[-1].strip()
-        else:
-            generated_code = full_response.strip()
-        if len(generated_code) > 2000:
-            generated_code = generated_code[:2000] + "\n// ... (truncated)"
-        return generated_code
-    except Exception as e:
-        return f"Error generating code: {e}"
-
-def gradio_interface(style_input, max_tokens, temperature, top_p, use_sampling):
-    instruction = "Generate Strudel.cc JavaScript code given style tags and a prompt, without commentary or markdown."
-    return generate_code(
-        instruction=instruction,
-        style_input=style_input,
-        max_new_tokens=int(max_tokens),
-        temperature=temperature,
-        top_p=top_p,
-        do_sample=use_sampling
-    )
-
-# Build the Blocks UI
-load_model()
-with gr.Blocks(title="Strudel.cc Code Generator", theme=gr.themes.Soft()) as demo:
-    gr.Markdown("""
-    # 🎵 Strudel.cc Code Generator
-    
-    Generate Strudel.cc JavaScript code based on style descriptions or musical concepts.
-    
-    **Model:** robertnetwork/strudel-small (Fine-tuned with LoRA)
-    """)
-    with gr.Row():
-        with gr.Column(scale=1, min_width=300):
-            style_input = gr.Textbox(label="Style/Prompt", placeholder="e.g., electronic, ambient…", lines=3, value="electronic")
-            with gr.Accordion("Generation Settings", open=False):
-                max_tokens   = gr.Slider(50, 256, value=128, step=32, label="Max New Tokens")
-                temperature  = gr.Slider(0.1, 2.0, value=0.7, step=0.1, label="Temperature")
-                top_p        = gr.Slider(0.1, 1.0, value=0.9, step=0.1, label="Top-p")
-                use_sampling = gr.Checkbox(label="Use Sampling", value=True)
-            generate_btn = gr.Button("🎵 Generate Code", variant="primary", size="lg")
-        with gr.Column(scale=2):
-            output_code = gr.Code(label="Generated Strudel.cc Code", language="javascript", lines=20)
-
-    gr.Markdown("### 🎯 Example Prompts")
-    examples = gr.Examples(
-        examples=[["electronic"], ["ambient drone"], ["techno beat"], ["drum and bass"], ["minimalist"], ["glitch hop"], ["house music"], ["experimental"]],
-        inputs=[style_input]
-    )
-
-    generate_btn.click(fn=gradio_interface, inputs=[style_input, max_tokens, temperature, top_p, use_sampling], outputs=[output_code])
-    style_input.submit (fn=gradio_interface, inputs=[style_input, max_tokens, temperature, top_p, use_sampling], outputs=[output_code])
-
-# ──────────────── LAUNCH WITH UVIORN + HTTPOOLS ────────────────
-if __name__ == "__main__":
-    # demo.queue()  # uncomment if you want background queuing support
-    uvicorn.run(
-        demo.app, 
-        host="0.0.0.0", 
-        port=7860,
-        http="httptools",     # ← switch off H11
-        ws="websockets"
-    )