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.
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
Add TFConvNextModel #15750
Add TFConvNextModel #15750
Changes from 38 commits
d2a0848
66fc8fa
583769c
1e0a589
c667d93
7aecfa9
222c465
835dbdb
d6f91b6
e92b6ce
e1fec88
30e4bcb
b0051ac
aeb14f7
aec69dc
6c0fae2
ee62db4
8c1d6a3
490adf8
fa49469
18f0b0a
acb6fa0
077ee25
8d56711
11b0683
fd0ca7f
98911a2
b30a8cc
2181d5b
cc98979
12e4505
eb49338
5e01b71
3bd1c92
3aefac7
908d0cf
d386cf8
15c916f
05b8273
d247441
bb8e6c2
3b5366d
d375214
49b35cd
d9b5079
4b4737f
2322a5f
61ae121
b568377
1259bf8
96c1ea4
0f98fb5
b197216
7dcd98a
95fffed
f8129a1
dab6866
9d6b8ad
69b5413
15c6814
e39c41b
98111f8
3e06942
bc46016
06e19cd
229a817
ad5d7e0
4dea175
0f8069d
f4292b4
8b99c8e
7819850
ba9484f
553bac5
d22e0cb
de00fb2
b2309fe
File filter
Filter by extension
Conversations
Jump to
Diff view
Diff view
There are no files selected for viewing
Original file line number | Diff line number | Diff line change | ||
---|---|---|---|---|
|
@@ -85,6 +85,7 @@ def __init__( | |||
is_encoder_decoder=False, | ||||
layer_scale_init_value=1e-6, | ||||
drop_path_rate=0.0, | ||||
image_size=224, | ||||
**kwargs | ||||
): | ||||
super().__init__(**kwargs) | ||||
|
@@ -99,3 +100,5 @@ def __init__( | |||
self.layer_norm_eps = layer_norm_eps | ||||
self.layer_scale_init_value = layer_scale_init_value | ||||
self.drop_path_rate = drop_path_rate | ||||
self.image_size = image_size | ||||
self.output_attentions = None | ||||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. This line causes PyTorch test failed
I understand why you set Error message on CircleCI
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Yeah happy to discuss this with others. There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. This line should not be present indeed. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This change is not related to this PR and the test makes it so that the config will override the value
output_attentions
passed if it'sFalse
which should not be the case. The test should beand this should really be in its own PR if it's fixing a bug.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Without this change, I don't think it'd be possible to deal with
output_attentions
argument in the TF model.What should be done for the PR then?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Just adding the context (if not clear yet):
booleans_processing()
assumes thatoutput_attentions
will be inkwargs
.This is true so far, as we always add
output_attentions
as an argument to the model arguments.ConvNextModel
is the 1st (?) one that doesn't hasoutput_attentions
argument.There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@sgugger @ydshieh just applied the change and added a comment to explain why it's required.