Andrew Gu a93ea617b5 [FSDP2] Required mesh_dim_names for HSDP (#137436) ... Two changes: 1. Require `mesh_dim_names` if using HSDP 2. Pass only the shard mesh to `fsdp_pre_all_gather` Change 1 is technically BC breaking, but it should not be hard to fix on the user side. Pull Request resolved: https://github.com/pytorch/pytorch/pull/137436 Approved by: https://github.com/weifengpy, https://github.com/wz337		2024-10-09 20:35:09 +00:00
..
test_fully_shard_autograd.py
test_fully_shard_clip_grad_norm_.py	[reland][dtensor] move DTensor to public namespace (#134203)	2024-09-08 17:08:40 +00:00
test_fully_shard_comm.py	[FSDP2] Added shard_placement_fn arg (#137496)	2024-10-09 19:13:32 +00:00
test_fully_shard_compile.py	[FSDP2] Added shard_placement_fn arg (#137496)	2024-10-09 19:13:32 +00:00
test_fully_shard_extensions.py	[FSDP2] Required mesh_dim_names for HSDP (#137436)	2024-10-09 20:35:09 +00:00
test_fully_shard_frozen.py
test_fully_shard_grad_scaler.py	Gradient scaler for DTensor (#132816)	2024-09-05 16:44:32 +00:00
test_fully_shard_init.py	[FSDP2] Required mesh_dim_names for HSDP (#137436)	2024-10-09 20:35:09 +00:00
test_fully_shard_logging.py
test_fully_shard_memory.py
test_fully_shard_mixed_precision.py	[FSDP2] Added shard_placement_fn arg (#137496)	2024-10-09 19:13:32 +00:00
test_fully_shard_overlap.py
test_fully_shard_state.py
test_fully_shard_state_dict.py	[FSDP2] Required mesh_dim_names for HSDP (#137436)	2024-10-09 20:35:09 +00:00
test_fully_shard_training.py	[FSDP2] Required mesh_dim_names for HSDP (#137436)	2024-10-09 20:35:09 +00:00