EDAnonSubmission/benchmark2

EditJudge-Bench Evaluation Code

This repository contains lightweight, anonymous evaluation utilities for the EditJudge-Bench dataset release. The code validates a local dataset snapshot, expands edit-level rows into verification triplets, and computes the AUROC metrics used to audit VLM-as-a-judge behaviour.

Dataset URL: https://huggingface.co/datasets/EDAnonSubmission/benchmark

Install

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Dataset Layout

The scripts expect a local dataset snapshot with:

benchmark.parquet
images/<sample_id>/before.jpg
images/<sample_id>/after.jpg

Each parquet row is one edit pair. A positive verification triplet is created from instruction_pos; negative triplets are created from instruction_neg_list and instruction_neg_types.

Validate the Dataset

python scripts/validate_dataset.py \
  --dataset-root /path/to/benchmark

The validator checks row count, image coverage, negative-instruction alignment, edit-type balance, and that image paths are portable and repo-relative.

Prediction Schema

Prediction files may be CSV, JSONL, JSON, or Parquet. The simplest schema is:

sample_id,example_type,negative_index,score

For positives, set example_type=positive and negative_index=-1. For negatives, set example_type=negative and use the index in instruction_neg_list.

The scripts also accept parquet_row_index instead of sample_id, or an expanded table that already contains label, edit_type, negative_type, and score.

Compute Main Metrics

python scripts/compute_benchpress_metrics.py \
  --dataset-root /path/to/benchmark \
  --predictions /path/to/predictions.parquet \
  --out-dir outputs/judge_name

Outputs:

• overall_metrics.csv: global AUROC and macro edit-type AUROC.
• per_edit_type_auc.csv: AUROC for each edit type.

Negative-Type Breakdown

python scripts/compute_negative_type_breakdown.py \
  --dataset-root /path/to/benchmark \
  --predictions /path/to/predictions.parquet \
  --out-dir outputs/judge_name

Outputs:

• per_negative_type_auc.csv
• semantic_vs_noedit_summary.csv

Ground-Truth Salience Tables

python scripts/make_salience_tables.py \
  --dataset-root /path/to/benchmark \
  --predictions /path/to/predictions.parquet \
  --out-dir outputs/judge_name \
  --bins 5

This conditions no-edit rejection AUROC on saved Blender parameters such as shape delta, articulation delta, scale delta, movement distance, rotation angle, lighting magnitude, and camera changes.

Notes

This code release evaluates judge scores; it does not run VLM inference. The paper's model-specific prompts and inference wrappers are implementation details around producing the prediction files consumed here.