SomeAI.org
  • Hot AI Tools
  • New AI Tools
  • AI Category
SomeAI.org
SomeAI.org

Discover 10,000+ free AI tools instantly. No login required.

About

  • Blog

© 2025 • SomeAI.org All rights reserved.

  • Privacy Policy
  • Terms of Service
Home
Model Benchmarking
OR-Bench Leaderboard

OR-Bench Leaderboard

Measure over-refusal in LLMs using OR-Bench

You May Also Like

View All
🏅

LLM HALLUCINATIONS TOOL

Evaluate AI-generated results for accuracy

0
🚀

EdgeTA

Retrain models for new data at edge devices

1
📊

MEDIC Benchmark

View and compare language model evaluations

8
🔀

mergekit-gui

Merge machine learning models using a YAML configuration file

271
🐠

PaddleOCRModelConverter

Convert PaddleOCR models to ONNX format

3
📈

Building And Deploying A Machine Learning Models Using Gradio Application

Predict customer churn based on input details

2
✂

MTEM Pruner

Multilingual Text Embedding Model Pruner

9
📏

Cetvel

Pergel: A Unified Benchmark for Evaluating Turkish LLMs

16
🏛

CaselawQA leaderboard (WIP)

Browse and submit evaluations for CaselawQA benchmarks

4
📉

Testmax

Download a TriplaneGaussian model checkpoint

0
🚀

Can You Run It? LLM version

Determine GPU requirements for large language models

950
🐠

WebGPU Embedding Benchmark

Measure BERT model performance using WASM and WebGPU

0

What is OR-Bench Leaderboard ?

OR-Bench Leaderboard is a tool designed to measure and benchmark over-refusal in Large Language Models (LLMs). It provides a standardized framework to evaluate when and how models refuse to answer questions or generate text. This leaderboard helps researchers and developers understand the limitations and safety mechanisms of LLMs by comparing their performance across different scenarios.

Features

• Model Comparison: Allows users to compare multiple models based on their refusal patterns.
• Refusal Trigger Evaluation: Tests models against a curated set of triggers to assess their refusal thresholds.
• Metric Aggregation: Provides aggregated metrics such as refusal rates and response patterns.
• Result Sharing: Enables sharing of benchmark results for community collaboration.
• Documented Methodology: Offers transparent documentation of evaluation methods and datasets.

How to use OR-Bench Leaderboard ?

  1. Visit the OR-Bench Leaderboard website: Access the platform to explore benchmarked models.
  2. Select a model: Choose from the list of available LLMs to view its performance.
  3. Input refusal triggers: Enter specific prompts or scenarios to test the model's refusal behavior.
  4. Run the benchmark: Execute the evaluation to generate results.
  5. Analyze metrics: Review refusal rates, response patterns, and other key metrics.
  6. Compare results: Use the leaderboard to compare results across different models.

Frequently Asked Questions

1. What is the purpose of OR-Bench Leaderboard?
The purpose of OR-Bench Leaderboard is to provide a standardized way to measure and compare over-refusal behaviors in LLMs, helping to identify models with balanced safety and utility.

2. Why is measuring over-refusal important?
Measuring over-refusal is important because it helps assess a model's ability to avoid harmful or inappropriate responses while still providing useful answers.

3. How can I interpret the results from OR-Bench Leaderboard?
Results show how often a model refuses to answer and under what conditions. Lower refusal rates may indicate a model that is more permissive, while higher rates suggest stricter safety mechanisms.

Recommended Category

View All
🚫

Detect harmful or offensive content in images

❓

Visual QA

📊

Data Visualization

❓

Question Answering

👤

Face Recognition

✨

Restore an old photo

🎮

Game AI

🌜

Transform a daytime scene into a night scene

🎧

Enhance audio quality

🎭

Character Animation

⭐

Recommendation Systems

😊

Sentiment Analysis

🗣️

Voice Cloning

🖼️

Image Captioning

🗂️

Dataset Creation