Implementation:Open compass VLMEvalKit VLRewardBench

Field	Value
source	VLMEvalKit
domain	Vision, Benchmarking, Reward Model Evaluation

Overview

Benchmark dataset implementation for VL-RewardBench evaluation in VLMEvalKit.

Description

VLRewardBench inherits from ImageBaseDataset and implements the VL-RewardBench benchmark for evaluating visual-language reward models. The TYPE field is set to 'VQA'. It handles multi-image prompts and evaluates model preference judgments.

Usage

Registered in vlmeval/dataset/__init__.py and invoked through build_dataset() by benchmark name.

Code Reference

Source: vlmeval/dataset/vl_rewardbench.py, Lines: L1-173
Import: from vlmeval.dataset.vl_rewardbench import VLRewardBench

Signature:

class VLRewardBench(ImageBaseDataset):
    TYPE = 'VQA'
    DATASET_URL = {...}
    DATASET_MD5 = {...}
    ...

I/O Contract

Direction	Description
Inputs	TSV dataset file with image/video paths and questions
Outputs	Evaluation results DataFrame with scores per category

Usage Examples

from vlmeval.dataset import build_dataset
dataset = build_dataset('VL-RewardBench')

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment