theluyuan/DiffSynth-Studio
1
0
Fork 0
mirror of https://github.com/modelscope/DiffSynth-Studio.git synced 2026-03-19 06:48:12 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #1272 from modelscope/zero3-fix

Browse Source 
Support DeepSpeed ZeRO 3
This commit is contained in:
Zhongjie Duan
2026-02-06 16:33:12 +08:00
committed by GitHub
parent abdf66d09e b0bf78e915
commit 1b47e1dc22
26 changed files with 353 additions and 188 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
diffsynth/utils/xfuser/xdit_context_parallel.py
View File

@@ -9,6 +9,7 @@ from xfuser.core.long_ctx_attention import xFuserLongContextAttention
from ... import IS_NPU_AVAILABLE
from ...core.device import parse_nccl_backend, parse_device_type
from ...core.gradient import gradient_checkpoint_forward
def initialize_usp(device_type):
@@ -87,11 +88,6 @@ def usp_dit_forward(self,
self.freqs[1][:h].view(1, h, 1, -1).expand(f, h, w, -1),
self.freqs[2][:w].view(1, 1, w, -1).expand(f, h, w, -1)
], dim=-1).reshape(f * h * w, 1, -1).to(x.device)
def create_custom_forward(module):
def custom_forward(*inputs):
return module(*inputs)
return custom_forward
# Context Parallel
chunks = torch.chunk(x, get_sequence_parallel_world_size(), dim=1)
@@ -100,20 +96,13 @@ def usp_dit_forward(self,
x = chunks[get_sequence_parallel_rank()]
for block in self.blocks:
if self.training and use_gradient_checkpointing:
if use_gradient_checkpointing_offload:
with torch.autograd.graph.save_on_cpu():
x = torch.utils.checkpoint.checkpoint(
create_custom_forward(block),
x, context, t_mod, freqs,
use_reentrant=False,
)
else:
x = torch.utils.checkpoint.checkpoint(
create_custom_forward(block),
x, context, t_mod, freqs,
use_reentrant=False,
)
if self.training:
x = gradient_checkpoint_forward(
block,
use_gradient_checkpointing,
use_gradient_checkpointing_offload,
x, context, t_mod, freqs
)
else:
x = block(x, context, t_mod, freqs)