Skip to content
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

PegasusX add _no_split_modules #25933

Merged
merged 7 commits into from
Sep 5, 2023
Merged

PegasusX add _no_split_modules #25933

Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
651d434
no_split_modules
andreeahedes Sep 1, 2023
1e5ffb1
no_split_modules
andreeahedes Sep 1, 2023
80c1800
inputs_embeds+pos same device
andreeahedes Sep 3, 2023
8facde2
Merge branch 'pegasus-x-no-split-modules' of github.com:andreeahedes/…
andreeahedes Sep 3, 2023
48c7d8a
update _no_split_modules
andreeahedes Sep 3, 2023
43e8623
update _no_split_modules
andreeahedes Sep 3, 2023
cfa9296
Merge branch 'pegasus-x-no-split-modules' of github.com:andreeahedes/…
andreeahedes Sep 3, 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
3 changes: 3 additions & 0 deletions src/transformers/models/pegasus_x/modeling_pegasus_x.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -769,6 +769,7 @@ class PegasusXPreTrainedModel(PreTrainedModel):
config_class = PegasusXConfig
base_model_prefix = "model"
supports_gradient_checkpointing = True
_no_split_modules = [r"PegasusXEncoderLayer", r"PegasusXDecoderLayer"]

def _init_weights(self, module):
std = self.config.init_std
Expand Down Expand Up @@ -1299,6 +1300,8 @@ def forward(
# embed positions
positions = self.embed_positions(inputs_embeds, past_key_values_length)

positions = positions.to(inputs_embeds.device)

SunMarc marked this conversation as resolved.
Show resolved Hide resolved
hidden_states = inputs_embeds + positions

hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
Expand Down