Delete app.py

app.py

@@ -1,121 +0,0 @@
-import torch
-from transformers import AutoTokenizer, AutoModel
-from huggingface_hub import HfApi, HfFolder, create_repo, upload_folder
-import os
-import gradio as gr
-
-def load_and_reupload_model(model_name, new_repo_id, hf_token, max_shard_size="1.5GB"):
-    """
-    Loads a model and tokenizer, saves the model in smaller shards,
-    and uploads them to a new private repository on the Hugging Face Hub.
-
-    Args:
-        model_name (str): The name of the model to load from the Hugging Face Hub.
-        new_repo_id (str): The ID for the new private repository (e.g., "your-username/your-repo-name").
-        hf_token (str): Your Hugging Face API token with write permissions.
-        max_shard_size (str, optional): The maximum size of each model shard. Defaults to "1GB".
-    
-    Returns:
-        A string log of the process.
-    """
-    log_output = []
-    try:
-        # --- Validate inputs ---
-        if not all([model_name, new_repo_id, hf_token]):
-            return "❌ Error: All fields (Model Name, New Repo ID, and HF Token) are required."
-
-        # --- 1. Log in to Hugging Face Hub ---
-        HfFolder.save_token(hf_token)
-        log_output.append("✅ Successfully saved Hugging Face token.")
-
-        # --- 2. Create a new private repository ---
-        create_repo(new_repo_id, private=True, exist_ok=True, token=hf_token)
-        log_output.append(f"✅ Successfully created or confirmed private repository: {new_repo_id}")
-
-        # --- 3. Load the tokenizer and model ---
-        log_output.append(f"🔄 Loading tokenizer for {model_name}...")
-        tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
-        log_output.append("✅ Tokenizer loaded successfully.")
-
-        log_output.append(f"🔄 Loading model {model_name}...")
-        device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-        model = AutoModel.from_pretrained(
-            model_name,
-            trust_remote_code=True,
-            use_safetensors=True,
-        ).to(device)
-
-        if device.type == 'cuda':
-            model = model.to(torch.bfloat16)
-
-        model.eval()
-        log_output.append(f"✅ Model loaded successfully to {device} and set to evaluation mode.")
-
-        # --- 4. Save the model and tokenizer locally with sharding ---
-        local_save_dir = new_repo_id.split('/')[-1]
-        os.makedirs(local_save_dir, exist_ok=True)
-
-        log_output.append(f"🔄 Saving model with max_shard_size='{max_shard_size}' to '{local_save_dir}'...")
-        model.save_pretrained(local_save_dir, max_shard_size=max_shard_size)
-        tokenizer.save_pretrained(local_save_dir)
-        log_output.append("✅ Model and tokenizer saved locally.")
-
-        # --- 5. Upload the sharded model and tokenizer to the new repo ---
-        log_output.append(f"🔄 Uploading files to {new_repo_id}...")
-        api = HfApi()
-        api.upload_folder(
-            folder_path=local_save_dir,
-            repo_id=new_repo_id,
-            repo_type="model",
-            token=hf_token
-        )
-        log_output.append(f"🚀 Successfully uploaded model and tokenizer to private repo: {new_repo_id}")
-
-    except Exception as e:
-        log_output.append(f"❌ An error occurred: {e}")
-    
-    return "\n".join(log_output)
-
-# --- Gradio Interface ---
-with gr.Blocks(theme="soft") as demo:
-    gr.Markdown(
-        """
-        # 🚀 Hugging Face Model Sharder & Re-Uploader
-        This application loads a model from the Hugging Face Hub, saves it locally into smaller shards (e.g., 1GB each), and then uploads it to a new private repository under your account.
-        """
-    )
-
-    with gr.Row():
-        with gr.Column(scale=2):
-            model_name_input = gr.Textbox(
-                label="Original Model Name",
-                value="strangervisionhf/deepseek-ocr-latest-transformers",
-                placeholder="e.g., 'strangervisionhf/deepseek-ocr-latest-transformers'"
-            )
-            new_repo_id_input = gr.Textbox(
-                label="New Private Repository ID",
-                placeholder="e.g., 'your-username/private-deepseek-ocr-sharded'"
-            )
-            hf_token_input = gr.Textbox(
-                label="Hugging Face Write Token",
-                type="password",
-                placeholder="Enter your Hugging Face token with write access"
-            )
-            run_button = gr.Button("Shard and Upload Model", variant="primary")
-        
-        with gr.Column(scale=3):
-            output_log = gr.Textbox(
-                label="Process Log",
-                lines=15,
-                interactive=False,
-                autoscroll=True
-            )
-
-    run_button.click(
-        fn=load_and_reupload_model,
-        inputs=[model_name_input, new_repo_id_input, hf_token_input],
-        outputs=output_log
-    )
-
-if __name__ == "__main__":
-    demo.launch()