init

app.py

@@ -24,7 +24,8 @@ def fine_tune_model(model_name, dataset_name, hub_id, num_epochs, batch_size, lr
     
         # Tokenize the dataset
         def tokenize_function(examples):
-            return tokenizer(examples['text'], padding="max_length", truncation=True)
+            max_length = 256
+            return tokenizer(examples['text'], padding=max_length, truncation=True)
     
         tokenized_datasets = dataset.map(tokenize_function, batched=True)
     
@@ -33,12 +34,12 @@ def fine_tune_model(model_name, dataset_name, hub_id, num_epochs, batch_size, lr
             output_dir='./results',
             eval_strategy="epoch",
             save_strategy='epoch', 
-            learning_rate=lr**-5,
+            learning_rate=lr*0.00001,
             per_device_train_batch_size=batch_size,
             per_device_eval_batch_size=batch_size, 
             num_train_epochs=num_epochs,
             weight_decay=0.01,
-            gradient_accumulation_steps=grad**-1,
+            gradient_accumulation_steps=grad*0.1,
             load_best_model_at_end=True,
             metric_for_best_model="accuracy",
             greater_is_better=True,
@@ -49,7 +50,7 @@ def fine_tune_model(model_name, dataset_name, hub_id, num_epochs, batch_size, lr
             fp16=True,
             lr_scheduler_type='cosine',
         )
-    
+        
         # Create Trainer
         trainer = Trainer(
             model=model,