Embedding Leaderboard
Upload a table to predict basalt source lithology, temperature, and pressure
Ask questions about air quality data with pre-built prompts or your own queries
Parse and highlight entities in an email thread
Retrieve news articles based on a query
Generate answers by querying text in uploaded documents
Display and explore model leaderboards and chat history
Submit model predictions and view leaderboard results
Use title and abstract to predict future academic impact
Calculate love compatibility using names
A benchmark for open-source multi-dialect Arabic ASR models
Explore Arabic NLP tools
Aligns the tokens of two sentences
The MTEB Leaderboard is a comprehensive platform designed for evaluating and comparing text embedding models. It enables users to select specific benchmarks and languages to assess the performance of various text embeddings. This tool is particularly useful for researchers and developers in the field of natural language processing (NLP) who need to understand how different models perform across diverse tasks and languages.
• Customizable Benchmarks: Choose from a wide range of evaluation benchmarks tailored to different NLP tasks.
• Multilingual Support: Evaluate embeddings across multiple languages, making it ideal for multilingual NLP studies.
• Model Comparison: Directly compare the performance of different embedding models on the same tasks.
• Automated Evaluation: Streamline the evaluation process with automated pipelines for selected benchmarks.
• Visualization Tools: Access detailed visualizations to better understand model performance and differences.
• Regular Updates: Stay current with the latest models and benchmarks through frequent updates.
1. What languages does MTEB Leaderboard support?
MTEB Leaderboard supports a wide range of languages, including English, Spanish, French, German, Chinese, and many others. The exact list of supported languages can be found on the platform.
2. How do I update my model on the leaderboard?
To update your model on the leaderboard, re-run the evaluation with the new model version and submit the results through the platform's interface.
3. Can I create custom benchmarks?
Currently, MTEB Leaderboard offers pre-defined benchmarks, but there are plans to introduce custom benchmark creation in future updates. For now, you can select from the available benchmarks that best fit your needs.