Mismatch Between `train_file_names` and `train_files` Parameters Causes Assertion Failure

[Tangkexian]

Great work! But when using the write_selected_data.py script, a mismatch in the number of --train_file_names and --train_files parameters results in an assertion failure.

Steps to Reproduce

1. Run the Data Selection Script as the README Instructions:

   python3 -m less.data_selection.write_selected_data \
   --target_task_names ${TARGET_TASK_NAMES} \
   --train_file_names flan_v2 cot dolly oasst1 \
   --train_files ../data/train/processed/dolly/dolly_data.jsonl ../data/train/processed/oasst1/oasst1_data.jsonl \
   --output_path $SELECTED_DATA_OUTPUT_PATH \
   --percentage 0.05

2. Observe the Assertion Failure:

   The script contains the following assertion:

   assert len(args.train_file_names) == len(args.train_files)

   In this example, --train_file_names has 4 names (flan_v2, cot, dolly, oasst1), while --train_files only provides 2 file paths (dolly_data.jsonl and oasst1_data.jsonl). This mismatch triggers the assertion, causing the script to terminate unexpectedly.