scripts/run_codebert_training.sh

#!/usr/bin/env bash
set -euo pipefail

DATA="${1:-data/sql_errors_dev.parquet}"
OUTPUT="${2:-models/codebert-cross-encoder}"
SAMPLES="${3:-}"

CMD=(python -m src.hf_train_codebert --data "${DATA}" --output-dir "${OUTPUT}")

if [[ -n "${SAMPLES}" ]]; then
  CMD+=(--max-samples "${SAMPLES}")
fi

echo "==> Training CodeBERT cross-encoder..."
"${CMD[@]}"

echo "==> Done. Model at ${OUTPUT}"