View NSQL Scores for Models
Track, rank and evaluate open LLMs and chatbots
GIFT-Eval: A Benchmark for General Time Series Forecasting
Convert a Stable Diffusion XL checkpoint to Diffusers and open a PR
Display and filter leaderboard models
Measure over-refusal in LLMs using OR-Bench
Predict customer churn based on input details
Evaluate reward models for math reasoning
Evaluate LLM over-refusal rates with OR-Bench
Display model benchmark results
View and submit LLM benchmark evaluations
Evaluate AI-generated results for accuracy
Create and upload a Hugging Face model card
DuckDB NSQL Leaderboard is a tool designed to track and compare the performance of different models using the NSQL (Normalized SQL) benchmarking framework. It provides a centralized platform to view and analyze NSQL scores, enabling users to evaluate and compare model performance efficiently.
What is NSQL in DuckDB?
NSQL (Normalized SQL) is a benchmarking framework used to evaluate the performance of SQL query engines. It provides a standardized way to measure and compare query execution times across different systems.
How do I interpret the NSQL scores?
Higher NSQL scores generally indicate better performance. Scores are calculated based on the execution time of a suite of SQL queries, with faster execution times resulting in higher scores.
Can I customize the leaderboard view?
Yes, you can customize the leaderboard by filtering, sorting, and selecting specific models to compare. This allows you to focus on the models and metrics that are most relevant to your needs.