Skip to content

Allow disabling torch 2_0 attention #3273

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 3 commits into from
Apr 28, 2023
Merged

Allow disabling torch 2_0 attention #3273

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
7b6d3f4
Allow disabling torch 2_0 attention
patrickvonplaten Apr 28, 2023
8564d25
make style
patrickvonplaten Apr 28, 2023
917211f
Update src/diffusers/models/attention.py
patrickvonplaten Apr 28, 2023
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
6 changes: 3 additions & 3 deletions src/diffusers/models/attention.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -71,6 +71,7 @@ def __init__(
self.proj_attn = nn.Linear(channels, channels, bias=True)

self._use_memory_efficient_attention_xformers = False
self._use_2_0_attn = True
self._attention_op = None

def reshape_heads_to_batch_dim(self, tensor, merge_head_and_batch=True):
Expand Down Expand Up @@ -142,9 +143,8 @@ def forward(self, hidden_states):

scale = 1 / math.sqrt(self.channels / self.num_heads)

use_torch_2_0_attn = (
hasattr(F, "scaled_dot_product_attention") and not self._use_memory_efficient_attention_xformers
)
_use_2_0_attn = self._use_2_0_attn and not self._use_memory_efficient_attention_xformers
use_torch_2_0_attn = hasattr(F, "scaled_dot_product_attention") and _use_2_0_attn

query_proj = self.reshape_heads_to_batch_dim(query_proj, merge_head_and_batch=not use_torch_2_0_attn)
key_proj = self.reshape_heads_to_batch_dim(key_proj, merge_head_and_batch=not use_torch_2_0_attn)
Expand Down