Skip to content

[ConvNeXT] Fix drop_path_rate #17280

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
May 17, 2022
Merged

[ConvNeXT] Fix drop_path_rate #17280

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
939cc1a
Fix drop_path_rate
NielsRogge May 16, 2022
35adae7
Fix TF's drop path rate
NielsRogge May 16, 2022
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: 4 additions & 4 deletions src/transformers/models/convnext/modeling_convnext.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -209,8 +209,9 @@ class ConvNextEncoder(nn.Module):
def __init__(self, config):
super().__init__()
self.stages = nn.ModuleList()
drop_path_rates = [x.item() for x in torch.linspace(0, config.drop_path_rate, sum(config.depths))]
cur = 0
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I've removed this (unclear) cur variable, and instead implement it similar to Ross' implementation in timm.

drop_path_rates = [
x.tolist() for x in torch.linspace(0, config.drop_path_rate, sum(config.depths)).split(config.depths)
]
prev_chs = config.hidden_sizes[0]
for i in range(config.num_stages):
out_chs = config.hidden_sizes[i]
Expand All @@ -220,10 +221,9 @@ def __init__(self, config):
out_channels=out_chs,
stride=2 if i > 0 else 1,
depth=config.depths[i],
drop_path_rates=drop_path_rates[cur],
drop_path_rates=drop_path_rates[i],
)
self.stages.append(stage)
cur += config.depths[i]
prev_chs = out_chs

def forward(
Expand Down
8 changes: 4 additions & 4 deletions src/transformers/models/convnext/modeling_tf_convnext.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -235,8 +235,9 @@ class TFConvNextEncoder(tf.keras.layers.Layer):
def __init__(self, config, **kwargs):
super().__init__(**kwargs)
self.stages = []
drop_path_rates = [x for x in tf.linspace(0.0, config.drop_path_rate, sum(config.depths))]
cur = 0
drop_path_rates = [
x.tolist() for x in tf.linspace(0, config.drop_path_rate, sum(config.depths)).split(config.depths)
]
prev_chs = config.hidden_sizes[0]
for i in range(config.num_stages):
out_chs = config.hidden_sizes[i]
Expand All @@ -246,11 +247,10 @@ def __init__(self, config, **kwargs):
out_channels=out_chs,
stride=2 if i > 0 else 1,
depth=config.depths[i],
drop_path_rates=drop_path_rates[cur],
drop_path_rates=drop_path_rates[i],
name=f"stages.{i}",
)
self.stages.append(stage)
cur += config.depths[i]
prev_chs = out_chs

def call(self, hidden_states, output_hidden_states=False, return_dict=True):
Expand Down