#!/bin/bash
# Batch BA validation script

set -e

# Configuration
SEQUENCES_DIR="${1:-data/raw}"
OUTPUT_DIR="${2:-data/processed}"
MODEL_NAME="${3:-depth-anything/DA3-LARGE}"
MAX_SAMPLES="${4:-1000}"

echo "Running BA validation on sequences in: $SEQUENCES_DIR"
echo "Output directory: $OUTPUT_DIR"
echo "Model: $MODEL_NAME"
echo "Max samples: $MAX_SAMPLES"

# Create output directory
mkdir -p "$OUTPUT_DIR"

# Run dataset building
python -m ylff.cli build-dataset \
    --sequences-dir "$SEQUENCES_DIR" \
    --model-name "$MODEL_NAME" \
    --output "$OUTPUT_DIR/training_set.pkl" \
    --max-samples "$MAX_SAMPLES"

echo "BA validation complete!"
echo "Training set saved to: $OUTPUT_DIR/training_set.pkl"