From d5ba471b98feca1d4772733d59db8aa90753b8b0 Mon Sep 17 00:00:00 2001 From: Christian Medina <37550954+cmedinasoriano@users.noreply.github.com> Date: Tue, 30 Jun 2026 19:54:50 -0400 Subject: [PATCH] fix: auto-split dataset if split files missing --- train-on-this-server.sh | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/train-on-this-server.sh b/train-on-this-server.sh index 3f80f9c..8c9a87f 100755 --- a/train-on-this-server.sh +++ b/train-on-this-server.sh @@ -60,8 +60,13 @@ if [ ! -f "dataset/combined_20k.jsonl" ]; then exit 1 fi -echo "Preparing dataset..." -python3 training/scripts/prepare_dataset.py --input dataset/combined_20k.jsonl --output training/data +echo "Checking for train.jsonl and test.jsonl..." +if [ ! -f "training/data/train.jsonl" ] || [ ! -f "training/data/test.jsonl" ]; then + echo "Split files not found, creating them..." + python3 training/scripts/prepare_dataset.py --input dataset/combined_20k.jsonl --output training/data +else + echo "Split files already exist, using them." +fi echo "" echo "=== Starting Training ==="