Ruslan Vasilev's picture
1 1

Ruslan Vasilev

artnitolog

AI & ML interests

None yet

Organizations

Posts 2

view post
Post
2501
Recently, we open-sourced YaFSDP, Yandex’s tool for efficient distributed training of LLMs.

Here are some of the key ideas used in YaFSDP to provide speedup and memory savings over FSDP:
• Allocate and utilize just two buffers throughout the transformer for all collected weights to circumvent the torch memory allocator;
• Gather small normalization layers at the beginning of the iteration and average the gradients only at the end;
• Move gradient division to the very end of the backward pass.

To learn more about how YaFSDP works, check out our latest blog post: https://medium.com/yandex/yafsdp-a-tool-for-faster-llm-training-and-optimized-gpu-utilization-is-no-632b7539f5b3
view post
Post
2057
Today we are introducing YaFSDP, Yandex’s tool for efficient distributed LLM training. YaFSDP can be used in conjunction with huggingface workflows and is up to 25% faster compared to FSDP.

Learn more here: https://github.com/yandex/YaFSDP

models

None public yet