Update app.py

app.py

@@ -97,6 +97,19 @@ if (runModel=='1'):
     test_dataset = IntentDataset(test_encodings, list(test_df['label']))
 
  
+    # Your repository name 
+    repo_name = "Reyad-Ahmmed/hf-data-timeframe" 
+    
+    
+    api_token = os.getenv("HF_API_TOKEN")  # Retrieve the API token from environment variable
+ 
+    if not api_token:
+        raise ValueError("API token not found. Please set the HF_API_TOKEN environment variable.")
+
+    # Create repository (if not already created)
+    api = HfApi()
+    create_repo(repo_id=repo_name, token=api_token, exist_ok=True)
+    
            
     # Create an instance of the custom loss function
     training_args = TrainingArguments(
@@ -109,6 +122,15 @@ if (runModel=='1'):
         logging_dir='./logs_' + modelNameToUse,
         logging_steps=10,
         evaluation_strategy="epoch",
+        
+    )
+
+    upload_folder( 
+        folder_path=training_args.output_dir, 
+        path_in_repo=f"{modelNameToUse}_results", 
+        repo_id=repo_name, 
+        token=api_token, 
+        commit_message="Upload training results" 
     )
 
     trainer = Trainer(
@@ -173,31 +195,29 @@ if (runModel=='1'):
     #model.save_pretrained('./'  + modelNameToUse + '_model')
     #tokenizer.save_pretrained('./' + modelNameToUse + '_tokenizer')
 
-    repo_name = "Reyad-Ahmmed/hf-data-timeframe"  # Replace with your repository name
-    api_token = os.getenv("HF_API_TOKEN")  # Replace with your actual API token
+    # Save the model and tokenizer locally
+    local_model_path = "./data-timeframe_model"
+    local_tokenizer_path = "./data-timeframe_tokenizer"
 
-    print("app token: ", api_token)
-    
-    api = HfApi() 
-    create_repo(repo_id=repo_name, token=api_token, exist_ok=True)
-    
-    model.save_pretrained("/data-timeframe_model")
-    tokenizer.save_pretrained("/data-timeframe_tokenizer")
+    model.save_pretrained(local_model_path)
+    tokenizer.save_pretrained(local_tokenizer_path)
 
     # Upload the model and tokenizer to the Hugging Face repository
-    upload_folder( 
-        folder_path="/data-timeframe_model", 
-        repo_id=repo_name, 
-        token=api_token, 
-        commit_message="Add fine-tuned model" 
-    ) 
-    
-    upload_folder( 
-        folder_path="/data-timeframe_tokenizer", 
-        repo_id=repo_name, 
-        token=api_token, 
-        commit_message="Add fine-tuned tokenizer" 
+    upload_folder(
+        folder_path=local_model_path,
+        path_in_repo="data-timeframe_model",
+        repo_id=repo_name,
+        token=api_token,
+        commit_message="Update fine-tuned model"
+    )
+    upload_folder(
+        folder_path=local_tokenizer_path,
+        path_in_repo="data-timeframe_tokenizer",
+        repo_id=repo_name,
+        token=api_token,
+        commit_message="Update fine-tuned tokenizer"
     )
+
     
 else:
     print('Load Pre-trained')