THU-KEG/OmniEvent
A modular deep learning toolkit for training, evaluating, and benchmarking event extraction models on diverse datasets.

Velocity · 7d
+0.3
★ / day
Trend
→steady
star history
OmniEvent is a unified event extraction framework built on PyTorch and HuggingFace Transformers. It provides standardized data processing pipelines, consistent evaluation metrics, and support for multiple model architectures across various event extraction datasets. The toolkit enables researchers to train custom models and perform comprehensive benchmarks with consistent preprocessing and output formats.