Skip to content

Fix enable_sequential_cpu_offload in test_kandinsky_combined #10324

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Dec 23, 2024
Merged

Fix enable_sequential_cpu_offload in test_kandinsky_combined #10324

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
f05f814
Fix enable_sequential_cpu_offload in test_kandinsky_combined
hlky Dec 20, 2024
d7f3b49
Merge branch 'main' into test-pipeline-kandinsky
sayakpaul Dec 21, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
18 changes: 9 additions & 9 deletions src/diffusers/pipelines/kandinsky/pipeline_kandinsky_combined.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -193,15 +193,15 @@ def __init__(
def enable_xformers_memory_efficient_attention(self, attention_op: Optional[Callable] = None):
self.decoder_pipe.enable_xformers_memory_efficient_attention(attention_op)

def enable_sequential_cpu_offload(self, gpu_id=0):
def enable_sequential_cpu_offload(self, gpu_id: Optional[int] = None, device: Union[torch.device, str] = "cuda"):
r"""
Offloads all models (`unet`, `text_encoder`, `vae`, and `safety checker` state dicts) to CPU using 🤗
Accelerate, significantly reducing memory usage. Models are moved to a `torch.device('meta')` and loaded on a
GPU only when their specific submodule's `forward` method is called. Offloading happens on a submodule basis.
Memory savings are higher than using `enable_model_cpu_offload`, but performance is lower.
"""
self.prior_pipe.enable_sequential_cpu_offload(gpu_id=gpu_id)
self.decoder_pipe.enable_sequential_cpu_offload(gpu_id=gpu_id)
self.prior_pipe.enable_sequential_cpu_offload(gpu_id=gpu_id, device=device)
self.decoder_pipe.enable_sequential_cpu_offload(gpu_id=gpu_id, device=device)

def progress_bar(self, iterable=None, total=None):
self.prior_pipe.progress_bar(iterable=iterable, total=total)
Expand Down Expand Up @@ -411,16 +411,16 @@ def __init__(
def enable_xformers_memory_efficient_attention(self, attention_op: Optional[Callable] = None):
self.decoder_pipe.enable_xformers_memory_efficient_attention(attention_op)

def enable_sequential_cpu_offload(self, gpu_id=0):
def enable_sequential_cpu_offload(self, gpu_id: Optional[int] = None, device: Union[torch.device, str] = "cuda"):
r"""
Offloads all models to CPU using accelerate, significantly reducing memory usage. When called, unet,
text_encoder, vae and safety checker have their state dicts saved to CPU and then are moved to a
`torch.device('meta') and loaded to GPU only when their specific submodule has its `forward` method called.
Note that offloading happens on a submodule basis. Memory savings are higher than with
`enable_model_cpu_offload`, but performance is lower.
"""
self.prior_pipe.enable_sequential_cpu_offload(gpu_id=gpu_id)
self.decoder_pipe.enable_sequential_cpu_offload(gpu_id=gpu_id)
self.prior_pipe.enable_sequential_cpu_offload(gpu_id=gpu_id, device=device)
self.decoder_pipe.enable_sequential_cpu_offload(gpu_id=gpu_id, device=device)

def progress_bar(self, iterable=None, total=None):
self.prior_pipe.progress_bar(iterable=iterable, total=total)
Expand Down Expand Up @@ -652,16 +652,16 @@ def __init__(
def enable_xformers_memory_efficient_attention(self, attention_op: Optional[Callable] = None):
self.decoder_pipe.enable_xformers_memory_efficient_attention(attention_op)

def enable_sequential_cpu_offload(self, gpu_id=0):
def enable_sequential_cpu_offload(self, gpu_id: Optional[int] = None, device: Union[torch.device, str] = "cuda"):
r"""
Offloads all models to CPU using accelerate, significantly reducing memory usage. When called, unet,
text_encoder, vae and safety checker have their state dicts saved to CPU and then are moved to a
`torch.device('meta') and loaded to GPU only when their specific submodule has its `forward` method called.
Note that offloading happens on a submodule basis. Memory savings are higher than with
`enable_model_cpu_offload`, but performance is lower.
"""
self.prior_pipe.enable_sequential_cpu_offload(gpu_id=gpu_id)
self.decoder_pipe.enable_sequential_cpu_offload(gpu_id=gpu_id)
self.prior_pipe.enable_sequential_cpu_offload(gpu_id=gpu_id, device=device)
self.decoder_pipe.enable_sequential_cpu_offload(gpu_id=gpu_id, device=device)

def progress_bar(self, iterable=None, total=None):
self.prior_pipe.progress_bar(iterable=iterable, total=total)
Expand Down
Loading