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

Jump to bottom

Add CvTONNX Config #1131

Merged
merged 3 commits into from
Jun 28, 2023
Merged

Add CvTONNX Config #1131

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
4774a57
CvTOnnx
rishabbala Jun 23, 2023
78dcf5f
Tests Passed
rishabbala Jun 23, 2023
1a6af8d
style
rishabbala Jun 26, 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
2 changes: 1 addition & 1 deletion optimum/bettertransformer/models/encoder_models.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -639,7 +639,7 @@ def forward(self, x, attn_mask, head_mask=None, output_attentions=None, *_):
attn_mask = torch.reshape(attn_mask, (attn_mask.shape[0], attn_mask.shape[-1]))
seqlen = attn_mask.shape[1]
lengths = torch.sum(~attn_mask, 1)
if not all([l == seqlen for l in lengths]):
if not all(l == seqlen for l in lengths):
x = torch._nested_tensor_from_mask(x, attn_mask)
attn_mask = None

Expand Down
5 changes: 5 additions & 0 deletions optimum/exporters/onnx/model_configs.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -550,6 +550,11 @@ def inputs(self) -> Dict[str, Dict[int, str]]:
return {"pixel_values": {0: "batch_size", 1: "num_channels", 2: "height", 3: "width"}}


class CvTOnnxConfig(ViTOnnxConfig):
DEFAULT_ONNX_OPSET = 13
ATOL_FOR_VALIDATION = 1e-2
Copy link
Member

Choose a reason for hiding this comment

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

This seems pretty high tbh, do you have any idea which op makes it fail?

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Are you talking about the tolerance or the opset version? Im not sure why lower tolerance doesn't work, but einsum causes an error for lower opset versions

Copy link
Member

Choose a reason for hiding this comment

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

I am talking about the tolerance, we might have a hidden bug if we need 1e-2, or we might not, but just wanted to raise the potential issue here.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Any way this can be checked?



class LevitOnnxConfig(ViTOnnxConfig):
pass

Expand Down
1 change: 1 addition & 0 deletions optimum/exporters/tasks.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -402,6 +402,7 @@ class TasksManager:
"image-classification",
onnx="ConvNextOnnxConfig",
),
"cvt": supported_tasks_mapping("feature-extraction", "image-classification", onnx="CvTOnnxConfig"),
"data2vec-text": supported_tasks_mapping(
"feature-extraction",
"fill-mask",
Expand Down
1 change: 1 addition & 0 deletions optimum/utils/normalized_config.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -205,6 +205,7 @@ class NormalizedConfigManager:
"bloom": NormalizedTextConfig.with_args(num_layers="n_layer"),
"camembert": NormalizedTextConfig,
"codegen": GPT2LikeNormalizedTextConfig,
"cvt": NormalizedVisionConfig,
"deberta": NormalizedTextConfig,
"deberta-v2": NormalizedTextConfig,
"deit": NormalizedVisionConfig,
Expand Down
1 change: 1 addition & 0 deletions tests/exporters/exporters_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -43,6 +43,7 @@
"clip": "hf-internal-testing/tiny-random-CLIPModel",
"convbert": "hf-internal-testing/tiny-random-ConvBertModel",
"codegen": "hf-internal-testing/tiny-random-CodeGenModel",
"cvt": "hf-internal-testing/tiny-random-CvTModel",
"data2vec-text": "hf-internal-testing/tiny-random-Data2VecTextModel",
"data2vec-vision": "hf-internal-testing/tiny-random-Data2VecVisionModel",
"data2vec-audio": "hf-internal-testing/tiny-random-Data2VecAudioModel",
Expand Down