pythonacceleratefsdp2pytorchafter-work
Accelerate 1.15 after work: FSDP2 activation memory knobs
Accelerate v1.15.0 (9 Sep 2026): FSDP2 activation checkpointing wraps the matched transformer layer (not each child); new fsdp_activation_checkpointing_offload to pinned CPU; tied-embedding fix on torch>=2.13; DTensor grad clip / prepare_model. Hook is Accelerate. Not investment advice.