Delete app.py

app.py

@@ -1,186 +0,0 @@
-import spaces
-import torch
-import tempfile
-import shutil
-import gradio as gr
-
-from transformers import (
-    AutoModel,
-    AutoModelForCausalLM,
-    AutoTokenizer,
-    AutoProcessor,
-)
-
-from huggingface_hub import create_repo, upload_folder
-from llmcompressor import oneshot
-from llmcompressor.modifiers.quantization import QuantizationModifier
-from llmcompressor.utils import dispatch_for_generation
-
-@spaces.GPU(duration=300)
-def fp8_dynamic_upload(
-    source_model,
-    target_repo,
-    hf_token,
-    max_new_tokens
-):
-    try:
-        if not source_model or not target_repo or not hf_token:
-            return "❌ Please fill all required fields."
-
-        logs = []
-        device = "cuda" if torch.cuda.is_available() else "cpu"
-
-        # -------------------------------------------------
-        # Load model + processor
-        # -------------------------------------------------
-        logs.append(f"🚀 Loading model: {source_model}")
-
-        processor = AutoProcessor.from_pretrained(
-            source_model,
-            trust_remote_code=True,
-            token=hf_token
-        )
-
-        model = AutoModelForCausalLM.from_pretrained(
-            source_model,
-            attn_implementation="flash_attention_2",
-            torch_dtype=torch.bfloat16,
-            device_map="auto",
-            trust_remote_code=True,
-            token=hf_token
-        ).eval()
-
-        logs.append("✅ Model loaded successfully")
-
-        # -------------------------------------------------
-        # Apply FP8_DYNAMIC Quantization
-        # -------------------------------------------------
-        logs.append("🔧 Applying FP8_DYNAMIC quantization")
-
-        recipe = QuantizationModifier(
-            targets="Linear",
-            scheme="FP8_DYNAMIC",
-            ignore=[
-                "lm_head",
-            ],
-        )
-
-        oneshot(model=model, recipe=recipe)
-
-        logs.append("✅ Quantization applied successfully")
-
-        # -------------------------------------------------
-        # Sanity Generation
-        # -------------------------------------------------
-        logs.append("🧠 Running sanity generation")
-
-        dispatch_for_generation(model)
-
-        inputs = processor(
-            text="Hello my name is",
-            return_tensors="pt"
-        ).to(device)
-
-        with torch.no_grad():
-            output = model.generate(
-                **inputs,
-                max_new_tokens=int(max_new_tokens)
-            )
-
-        sample_text = processor.decode(
-            output[0],
-            skip_special_tokens=True
-        )
-
-        logs.append("✅ Generation successful")
-        logs.append(f"Sample Output:\n{sample_text}")
-
-        # -------------------------------------------------
-        # Save compressed model
-        # -------------------------------------------------
-        tmp_dir = tempfile.mkdtemp()
-        logs.append("💾 Saving compressed model")
-
-        model.save_pretrained(
-            tmp_dir,
-            save_compressed=True
-        )
-
-        processor.save_pretrained(tmp_dir)
-
-        # -------------------------------------------------
-        # Create repo + Upload
-        # -------------------------------------------------
-        logs.append("☁ Creating private repo if not exists")
-
-        create_repo(
-            repo_id=target_repo,
-            token=hf_token,
-            private=True,
-            exist_ok=True
-        )
-
-        logs.append("⬆ Uploading to Hugging Face")
-
-        upload_folder(
-            repo_id=target_repo,
-            folder_path=tmp_dir,
-            token=hf_token
-        )
-
-        shutil.rmtree(tmp_dir)
-
-        logs.append("🎉 Upload completed successfully!")
-
-        return "\n\n".join(logs)
-
-    except Exception as e:
-        return f"❌ Error: {str(e)}"
-
-
-# -------------------------------------------------
-# Gradio UI
-# -------------------------------------------------
-with gr.Blocks(title="Dots OCR FP8_DYNAMIC Uploader") as app:
-
-    gr.Markdown("## 🔥 Dots OCR 1.5 → FP8_DYNAMIC → Hugging Face Upload")
-
-    source_model = gr.Textbox(
-        label="Model Source (HF Path)",
-        value="rednote-hilab/dots.ocr-1.5"
-    )
-
-    target_repo = gr.Textbox(
-        label="Target Repo (username/repo-name)",
-        placeholder="e.g. yourname/dots-ocr-fp8-dynamic"
-    )
-
-    hf_token = gr.Textbox(
-        label="Hugging Face Write Token",
-        type="password"
-    )
-
-    max_new_tokens = gr.Number(
-        label="Sanity Generation Max New Tokens",
-        value=20
-    )
-
-    run_btn = gr.Button("🚀 Quantize FP8_DYNAMIC & Upload")
-
-    output = gr.Textbox(
-        label="Status Log",
-        lines=22
-    )
-
-    run_btn.click(
-        fn=fp8_dynamic_upload,
-        inputs=[
-            source_model,
-            target_repo,
-            hf_token,
-            max_new_tokens
-        ],
-        outputs=output
-    )
-
-app.launch()