Upload scripts/filter_data.sh
Browse files- scripts/filter_data.sh +80 -0
scripts/filter_data.sh
ADDED
|
@@ -0,0 +1,80 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/bin/bash
|
| 2 |
+
# Data Filtering Pipeline for GUI-Shift
|
| 3 |
+
|
| 4 |
+
set -e
|
| 5 |
+
|
| 6 |
+
# Default values
|
| 7 |
+
INPUT_FILE=""
|
| 8 |
+
OUTPUT_FILE=""
|
| 9 |
+
MODEL_PATH="Qwen/Qwen2.5-VL-7B-Instruct"
|
| 10 |
+
NUM_GENERATIONS=8
|
| 11 |
+
TEMPERATURE=0.9
|
| 12 |
+
DEVICE="cuda"
|
| 13 |
+
SEED=42
|
| 14 |
+
|
| 15 |
+
# Parse arguments
|
| 16 |
+
while [[ $# -gt 0 ]]; do
|
| 17 |
+
case $1 in
|
| 18 |
+
--input_file)
|
| 19 |
+
INPUT_FILE="$2"
|
| 20 |
+
shift 2
|
| 21 |
+
;;
|
| 22 |
+
--output_file)
|
| 23 |
+
OUTPUT_FILE="$2"
|
| 24 |
+
shift 2
|
| 25 |
+
;;
|
| 26 |
+
--model_path)
|
| 27 |
+
MODEL_PATH="$2"
|
| 28 |
+
shift 2
|
| 29 |
+
;;
|
| 30 |
+
--num_generations)
|
| 31 |
+
NUM_GENERATIONS="$2"
|
| 32 |
+
shift 2
|
| 33 |
+
;;
|
| 34 |
+
--temperature)
|
| 35 |
+
TEMPERATURE="$2"
|
| 36 |
+
shift 2
|
| 37 |
+
;;
|
| 38 |
+
--device)
|
| 39 |
+
DEVICE="$2"
|
| 40 |
+
shift 2
|
| 41 |
+
;;
|
| 42 |
+
--seed)
|
| 43 |
+
SEED="$2"
|
| 44 |
+
shift 2
|
| 45 |
+
;;
|
| 46 |
+
*)
|
| 47 |
+
echo "Unknown option: $1"
|
| 48 |
+
exit 1
|
| 49 |
+
;;
|
| 50 |
+
esac
|
| 51 |
+
done
|
| 52 |
+
|
| 53 |
+
if [ -z "$INPUT_FILE" ] || [ -z "$OUTPUT_FILE" ]; then
|
| 54 |
+
echo "Error: --input_file and --output_file are required"
|
| 55 |
+
echo "Usage: bash scripts/filter_data.sh --input_file ./data/k1_transition.jsonl --output_file ./data/filtered/k1_transition_filtered.jsonl --model_path Qwen/Qwen2.5-VL-7B-Instruct"
|
| 56 |
+
exit 1
|
| 57 |
+
fi
|
| 58 |
+
|
| 59 |
+
echo "=== GUI-Shift Data Filtering ==="
|
| 60 |
+
echo "Input: $INPUT_FILE"
|
| 61 |
+
echo "Output: $OUTPUT_FILE"
|
| 62 |
+
echo "Model: $MODEL_PATH"
|
| 63 |
+
echo "N generations: $NUM_GENERATIONS"
|
| 64 |
+
echo "Temperature: $TEMPERATURE"
|
| 65 |
+
echo "Device: $DEVICE"
|
| 66 |
+
echo ""
|
| 67 |
+
|
| 68 |
+
mkdir -p "$(dirname "$OUTPUT_FILE")"
|
| 69 |
+
|
| 70 |
+
python src/filtering/filter_data.py \
|
| 71 |
+
--input_file "$INPUT_FILE" \
|
| 72 |
+
--output_file "$OUTPUT_FILE" \
|
| 73 |
+
--model_path "$MODEL_PATH" \
|
| 74 |
+
--num_generations "$NUM_GENERATIONS" \
|
| 75 |
+
--temperature "$TEMPERATURE" \
|
| 76 |
+
--device "$DEVICE" \
|
| 77 |
+
--seed "$SEED"
|
| 78 |
+
|
| 79 |
+
echo ""
|
| 80 |
+
echo "Filtering complete! Output: $OUTPUT_FILE"
|