diff --git a/README.md b/README.md
index 2f37e8f..17fadf2 100644
--- a/README.md
+++ b/README.md
@@ -99,8 +99,8 @@ The script `annotate_data/annotate.sh` does large-scale data annotation using a
 Example usage:
 ```python
 python select_training_data.py \
-    --input_base "datasets/Corpus-200B" \
-    --output_base "datasets/selected/Baseline-30B" \
+    "datasets/Corpus-200B" \
+    "datasets/selected/Baseline-30B" \
     --num_tokens 30000000000 \
     --do_sample \
     --num_proc 16