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

Autograd.grad support allow_unused argument #10251

Merged
merged 2 commits into from
May 11, 2023
Merged

Autograd.grad support allow_unused argument #10251

merged 2 commits into from
May 11, 2023

Conversation

pingzhuu
Copy link
Contributor

No description provided.

@pingzhuu pingzhuu requested review from BBuf and daquexian as code owners May 11, 2023 03:32
pingzhuu
pingzhuu commented May 11, 2023
View reviewed changes
python/oneflow/autograd/autograd.py
)
return tuple([Tensor(x) for x in in_grads])
return tuple([x for x in in_grads])
Copy link
Contributor Author

Choose a reason for hiding this comment

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

allow_unused=True 时Tensor(None) 会返回 nan

@github-actions
Copy link
Contributor

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/10251/

@github-actions
Copy link
Contributor

Speed stats: 
GPU Name: NVIDIA GeForce RTX 3080 Ti 

❌ OneFlow resnet50 time: 43.5ms (= 4352.6ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 61.6ms (= 6164.6ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.42 (= 61.6ms / 43.5ms)

OneFlow resnet50 time: 26.4ms (= 2637.3ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 37.7ms (= 3774.7ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.43 (= 37.7ms / 26.4ms)

OneFlow resnet50 time: 19.5ms (= 3904.2ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 36.6ms (= 7324.5ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.88 (= 36.6ms / 19.5ms)

OneFlow resnet50 time: 16.8ms (= 3353.6ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 30.7ms (= 6146.6ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.83 (= 30.7ms / 16.8ms)

OneFlow resnet50 time: 17.2ms (= 3445.7ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 28.7ms (= 5743.3ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.67 (= 28.7ms / 17.2ms)

OneFlow swin dataloader time: 0.202s (= 40.340s / 200, num_workers=1)
PyTorch swin dataloader time: 0.129s (= 25.703s / 200, num_workers=1)
Relative speed: 0.637 (= 0.129s / 0.202s)

OneFlow swin dataloader time: 0.056s (= 11.126s / 200, num_workers=4)
PyTorch swin dataloader time: 0.032s (= 6.498s / 200, num_workers=4)
Relative speed: 0.584 (= 0.032s / 0.056s)

OneFlow swin dataloader time: 0.029s (= 5.874s / 200, num_workers=8)
PyTorch swin dataloader time: 0.016s (= 3.296s / 200, num_workers=8)
Relative speed: 0.561 (= 0.016s / 0.029s)

❌ OneFlow resnet50 time: 47.4ms (= 4743.0ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 64.5ms (= 6448.0ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.36 (= 64.5ms / 47.4ms)

OneFlow resnet50 time: 31.0ms (= 3101.3ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 44.1ms (= 4412.4ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.42 (= 44.1ms / 31.0ms)

OneFlow resnet50 time: 24.3ms (= 4853.7ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 41.7ms (= 8345.1ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.72 (= 41.7ms / 24.3ms)

OneFlow resnet50 time: 22.5ms (= 4494.9ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 38.4ms (= 7687.7ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.71 (= 38.4ms / 22.5ms)

OneFlow resnet50 time: 21.4ms (= 4279.2ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 34.3ms (= 6851.5ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.60 (= 34.3ms / 21.4ms)

levi131
levi131 approved these changes May 11, 2023
View reviewed changes
Copy link
Contributor

@levi131 levi131 left a comment

Choose a reason for hiding this comment

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

LGTM

@mergify mergify bot merged commit f6de3a1 into master May 11, 2023
@mergify mergify bot deleted the autograd_support_allow_unused branch May 11, 2023 07:17
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@hjchen2 hjchen2 hjchen2 approved these changes

@levi131 levi131 levi131 approved these changes

@BBuf BBuf Awaiting requested review from BBuf

@daquexian daquexian Awaiting requested review from daquexian

Assignees
No one assigned
Labels
automerge enhancement test
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@pingzhuu @hjchen2 @levi131