GIFT-Eval: A Benchmark for General Time Series Forecasting
Measure over-refusal in LLMs using OR-Bench
Display leaderboard of language model evaluations
Quantize a model for faster inference
Display LLM benchmark leaderboard and info
Calculate memory usage for LLM models
Display leaderboard for earthquake intent classification models
Determine GPU requirements for large language models
Display and submit LLM benchmarks
Convert and upload model files for Stable Diffusion
Create demo spaces for models on Hugging Face
Track, rank and evaluate open LLMs and chatbots
View and submit LLM benchmark evaluations
GIFT-Eval is a benchmark platform designed for general time series forecasting. It provides a standardized framework to evaluate and compare the performance of various forecasting models across diverse time series datasets. The platform aims to foster research and development in time series analysis by offering a comprehensive leaderboard and analysis tools.
• Diverse Datasets: Includes a wide range of time series datasets from different domains. • Multiple Metrics: Evaluates forecasting models using various accuracy metrics. • Model Support: Compatible with popular time series forecasting models. • Leaderboard: Displays performance rankings of different models. • Open Source: Accessible for research and experimentation. • Comprehensive Documentation: Provides detailed guidelines and best practices.
What is the purpose of GIFT Eval?
GIFT-Eval is designed to provide a standardized benchmark for comparing time series forecasting models, enabling researchers and practitioners to evaluate model performance comprehensively.
How do I submit my model to GIFT Eval?
To submit your model, follow the platform's documentation to format your data and results correctly, then upload them through the provided interface.
Can I use GIFT Eval for my own datasets?
Yes, GIFT-Eval supports custom datasets. Simply format your data according to the platform's requirements and run the benchmarking process to evaluate your models.