Create app.py

app.py

@@ -0,0 +1,71 @@
+import gradio as gr
+import os
+import zipfile
+import glob
+from huggingface_hub import login, HfApi, create_repo
+
+def export_model_to_hf(hftoken, experiment_name, manual_epoch_number, logs_path, repoid, create_new_repo):
+    num_epochs = int(manual_epoch_number) if manual_epoch_number.isdigit() else None
+    
+    # Construct the weights path based on the provided number of epochs
+    if num_epochs is not None:
+        weights_path = f"/content/RVC/assets/weights/{experiment_name}_e{num_epochs}*"
+    else:
+        potential = f"/content/RVC/assets/weights/{experiment_name}.pth"
+        if os.path.exists(potential):
+            weights_path = f"/content/RVC/assets/weights/{experiment_name}"
+        else:
+            currentMax = 0
+            for r, _, f in os.walk("/content/RVC/assets/weights/"):
+                for name in f:
+                    if(name.endswith(".pth") and (name != experiment_name + ".pth")):
+                        if(name.find(experiment_name) == -1):
+                            continue
+                        pot = name.split('_')
+                        ep = pot[len(pot) - 2][1:]
+                        if not ep.isdecimal():
+                            continue
+                        ep = int(ep)
+                        if ep > currentMax:
+                            currentMax = ep
+            num_epochs = currentMax
+            weights_path = f"/content/RVC/assets/weights/{experiment_name}_e{num_epochs}*"
+    
+    weights_files = glob.glob(weights_path + ".pth")
+    if weights_files and any(glob_result := glob.glob(logs_path)):
+        log_file = glob_result[0]
+        output_folder = "/content/toHF"
+        os.makedirs(output_folder, exist_ok=True)
+        output_zip_path = f"{output_folder}/{experiment_name}.zip"
+        with zipfile.ZipFile(output_zip_path, 'w') as zipf:
+            for weights_file in weights_files:
+                zipf.write(weights_file, os.path.basename(weights_file))
+            zipf.write(log_file, os.path.basename(log_file))
+        
+        login(token=hftoken)
+        if create_new_repo:
+            create_repo(repoid)
+        
+        api = HfApi()
+        api.upload_folder(folder_path=output_folder, repo_id=repoid, repo_type="model")
+        
+        return f"Model uploaded successfully to {repoid}"
+    else:
+        return "Couldn't find your model files. Check the found file results above. (Did you run Index Training?)"
+
+with gr.Blocks() as demo:
+    gr.Markdown("<small>Export Finished Model to HuggingFace<br>[click this to get HF token](https://huggingface.co/settings/tokens)</small>")
+    
+    hftoken = gr.Textbox(label="HuggingFace Token (set Role to 'write')", type="password")
+    experiment_name = gr.Textbox(label="Experiment Name", value="rewrite")
+    manual_epoch_number = gr.Textbox(label="Manual Epoch Number (leave blank for auto-detect)", value="")
+    logs_path = gr.Textbox(label="Logs Path", value="/content/RVC/logs/rewrite/added_IVF37_Flat_nprobe_1_rewrite_v2.index")
+    repoid = gr.Textbox(label="HuggingFace Repository ID", value="Hev832/rewrite-sonic")
+    create_new_repo = gr.Checkbox(label="Create New Repository", value=True)
+    
+    output = gr.Textbox(label="Output")
+    
+    btn = gr.Button("Export Model")
+    btn.click(export_model_to_hf, inputs=[hftoken, experiment_name, manual_epoch_number, logs_path, repoid, create_new_repo], outputs=output)
+
+demo.launch()