Suggest an alternative to

Megatron-DeepSpeed

Ongoing research training transformer language models at scale, including: BERT & GPT-2

Why do you think that https://github.com/EleutherAI/gpt-neox is a good alternative to Megatron-DeepSpeed

A URL to the alternative repo (e.g. GitHub, GitLab)

Here you can share your experience with the project you are suggesting or its comparison with Megatron-DeepSpeed. Optional.

A valid email to send you a verification link when necessary or log in.