veRL is a flexible and efficient RL framework for LLMs.

A native PyTorch Library for large model training.

Ongoing research training transformer models at scale.

DeepSpeed version of NVIDIA's Megatron-LM that adds additional support for several features such as MoE model training, Curriculum Learning, 3D Parallelism, and others.

Making large AI models cheaper, faster, and more accessible.

A Native-PyTorch Library for LLM Fine-tuning.