infrastructure
FSDP2 activation memory reduced for long sequences in v1.15.0
FSDP2 activation checkpointing now wraps the entire transformer layer, reducing saved activations.
1 sources · 1 verified claimsOne evolving timeline of verified developments, without duplicating the underlying truth across pages.
FSDP2 activation checkpointing now wraps the entire transformer layer, reducing saved activations.
1 sources · 1 verified claims