debugging automatic job recovery

vms/ui/app_ui.py

@@ -434,14 +434,12 @@ class AppUI:
             training_preset = list(TRAINING_PRESETS.keys())[0]
             logger.warning(f"Invalid training preset '{training_preset}', using default: {training_preset}")
         
-        # Rest of the function remains unchanged
         lora_rank_val = ui_state.get("lora_rank", DEFAULT_LORA_RANK_STR)
         lora_alpha_val = ui_state.get("lora_alpha", DEFAULT_LORA_ALPHA_STR)
         batch_size_val = int(ui_state.get("batch_size", DEFAULT_BATCH_SIZE))
         learning_rate_val = float(ui_state.get("learning_rate", DEFAULT_LEARNING_RATE))
         save_iterations_val = int(ui_state.get("save_iterations", DEFAULT_SAVE_CHECKPOINT_EVERY_N_STEPS))
         
-        # Update for new UI components
         num_gpus_val = int(ui_state.get("num_gpus", DEFAULT_NUM_GPUS))
         
         # Calculate recommended precomputation items based on video count

vms/ui/project/services/previewing.py

@@ -520,7 +520,6 @@ class PreviewingService:
             log_fn("Starting video generation...")
             start_time.record()
             
-            # Fix for Issue #2: The pipe() expected list rather than float
             # Make sure negative_prompt is a list or None
             neg_prompt = [negative_prompt] if negative_prompt else None
             

vms/ui/project/services/training.py

@@ -1151,7 +1151,7 @@ class TrainingService:
                     # Use the internal model_type for the actual training
                     # But keep model_type_display for the UI
                     result = self.start_training(
-                        model_type=model_internal_type,
+                        model_type=model_type_internal,
                         lora_rank=params.get('lora_rank', DEFAULT_LORA_RANK_STR),
                         lora_alpha=params.get('lora_alpha', DEFAULT_LORA_ALPHA_STR),
                         train_size=params.get('train_steps', DEFAULT_NB_TRAINING_STEPS),