Skip to content

Fix LTX Video skip_layer_mask device mismatch under mmgp offloading #2015

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

Open
Jnalley123 wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Fix LTX Video skip_layer_mask device mismatch under mmgp offloading #2015

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
8 changes: 6 additions & 2 deletions models/ltx_video/models/transformers/attention.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -350,7 +350,7 @@ def forward(
skip_layer_mask is not None
and skip_layer_strategy == SkipLayerStrategy.TransformerBlock
):
skip_layer_mask = skip_layer_mask.view(-1, 1, 1)
skip_layer_mask = skip_layer_mask.view(-1, 1, 1).to(hidden_states.device)
hidden_states = hidden_states * skip_layer_mask + original_hidden_states * (
1.0 - skip_layer_mask
)
Expand Down Expand Up @@ -1014,7 +1014,11 @@ def __call__(
)

if skip_layer_mask is not None:
skip_layer_mask = skip_layer_mask.reshape(batch_size, 1, 1)
# The mask is created on the transformer's device, which is cpu when
# mmgp offloading is active, so follow the activations' device here.
skip_layer_mask = skip_layer_mask.reshape(batch_size, 1, 1).to(
hidden_states.device
)

if (attention_mask is not None) and (not attn.use_tpu_flash_attention):
attention_mask = attn.prepare_attention_mask(
Expand Down