T5X is a modular, composable, research-friendly framework for high-performance, configurable, self-service training, evaluation, and inference of sequence models (starting with language) at many scales. It is essentially a new and improved implementation of the [T5 codebase](https://github.com/google-research/text-to-text-transfer-transformer) (based on [Mesh TensorFlow](https://github.com/tensorflow/mesh)) in [JAX](https://github.com/google/jax) and [Flax](https://github.com/google/flax). To learn more, see the [T5X Paper](https://arxiv.org/abs/2203.17189).