A fast gradient checkpointing strategy for training with memory-efficient attention (e.g., FlashAttention).
Project description
FastCkpt: fast gradient checkpoint for training with FlashAttention
This is a test repo in development.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
fastckpt-0.0.3.tar.gz
(12.5 kB
view hashes)
Built Distribution
fastckpt-0.0.3-py3-none-any.whl
(11.9 kB
view hashes)