Update app.py

app.py

@@ -99,7 +99,7 @@ custom_data = [
 dataset_custom = load_dataset("json", data_files={"train": custom_data})
 
 # Merge with OpenWebText dataset
-dataset = load_dataset("Skylion007/openwebtext", split="train[:10%]")  # Load 5% to avoid streaming issues
+dataset = load_dataset("Skylion007/openwebtext", split="train[:50%]")  # Load 5% to avoid streaming issues
 
 # Tokenization function
 def tokenize_function(examples):