AuralGuard / scripts /run_ablation_template.sh
AyoPrince's picture
Upload folder using huggingface_hub
127b976 verified
Raw History Blame Contribute Delete
1.12 kB
#!/usr/bin/env bash
set -euo pipefail
# Edit these paths before running.
TRAIN_CSV="data/metadata/train.csv"
VAL_CSV="data/metadata/val.csv"
TEST_CSV="data/metadata/test.csv"
AASIST_ROOT="external/aasist"
AASIST_CONFIG="external/aasist/config/AASIST.conf"
# 1) Full unified model. The current starter trains all heads by default.
python -m src.train \
--train-csv "$TRAIN_CSV" \
--val-csv "$VAL_CSV" \
--aasist-root "$AASIST_ROOT" \
--aasist-config "$AASIST_CONFIG" \
--epochs 10 \
--batch-size 8 \
--out-dir results/full_unified
python -m src.evaluate \
--csv "$TEST_CSV" \
--checkpoint results/full_unified/best.pt \
--aasist-root "$AASIST_ROOT" \
--aasist-config "$AASIST_CONFIG" \
--out-csv results/full_unified/test_predictions.csv
# 2) Partial/localization evaluation if your CSV has start_fake and end_fake.
python -m src.evaluate_localization \
--csv "$TEST_CSV" \
--checkpoint results/full_unified/best.pt \
--aasist-root "$AASIST_ROOT" \
--aasist-config "$AASIST_CONFIG" \
--out-csv results/full_unified/localization_predictions.csv