GPU-optimized library for training transformer models at scale
NVIDIA's reference implementation for distributed transformer training with advanced parallelism strategies. Includes Megatron Core library and pre-configured training scripts for large language models.
Sign in to leave a comment.
No comments yet.
