dev

app.py

@@ -55,7 +55,7 @@ def fine_tune_model(model, dataset_name, hub_id, api_key, num_epochs, batch_size
                  
         # Set training arguments
         training_args = TrainingArguments(
-            torch_empty_cache_steps=1000, 
+            torch_empty_cache_steps=100, 
             overwrite_output_dir=True, 
             output_dir='/data/results',
             eval_strategy="steps",  # Change this to steps
@@ -71,12 +71,12 @@ def fine_tune_model(model, dataset_name, hub_id, api_key, num_epochs, batch_size
             metric_for_best_model="accuracy",
             greater_is_better=True,
             logging_dir='/data/logs',
-            logging_steps=2000,
+            logging_steps=300,
             #push_to_hub=True,
             hub_model_id=hub_id.strip(),
             fp16=True,
             #lr_scheduler_type='cosine',
-            save_steps=2000,  # Save checkpoint every 500 steps
+            save_steps=300,  # Save checkpoint every 500 steps
             save_total_limit=3, 
         )
         # Check if a checkpoint exists and load it