diff --git a/llm_training.py b/llm_training.py
index d7ec619..bf2e35c 100644
--- a/llm_training.py
+++ b/llm_training.py
@@ -60,7 +60,7 @@ def load_across_gpus(gpu_ids, batch_size, seq_length, epochs, learning_rate, cal
     inputs = {key: value.to(device) for key, value in inputs.items()}
     labels = inputs['input_ids']
 
-    # Use DataParallel to wrap the model for multi-GGPU usage
+    # Use DataParallel to wrap the model for multi-GPU usage
     model = DataParallel(model, device_ids=gpu_ids).to(device)
 
     initialize_nvml()