Support input check for pool operator #10532

Dmovic · 2024-05-24T02:43:00Z

related issues:

closed #10523
closed #10522
closed #10521

closed #10512
closed #10511
closed #10510
closed #10509
closed #10508
closed #10507
closed #10506
closed #10505
closed #10504
closed #10503
closed #10502
closed #10501
closed #10500

CLAassistant · 2024-05-24T02:43:05Z

All committers have signed the CLA.

github-actions · 2024-05-24T02:44:10Z

Code got formatted by CI. Please request CI again if you still want to have this PR merged. If the PR is from a forked repo, please download the patch files from the GitHub Actions web page and apply them locally.

github-actions · 2024-05-28T08:16:25Z

Code got formatted by CI. Please request CI again if you still want to have this PR merged. If the PR is from a forked repo, please download the patch files from the GitHub Actions web page and apply them locally.

github-actions · 2024-05-28T08:32:41Z

Code got formatted by CI. Please request CI again if you still want to have this PR merged. If the PR is from a forked repo, please download the patch files from the GitHub Actions web page and apply them locally.

…oneflow into support_input_check

github-actions · 2024-05-29T08:11:02Z

Code got formatted by CI. Please request CI again if you still want to have this PR merged. If the PR is from a forked repo, please download the patch files from the GitHub Actions web page and apply them locally.

github-actions · 2024-05-30T03:00:33Z

Code got formatted by CI. Please request CI again if you still want to have this PR merged. If the PR is from a forked repo, please download the patch files from the GitHub Actions web page and apply them locally.

github-actions · 2024-05-30T03:38:36Z

View latest API docs preview at: https://oneflow-staging.oss-cn-beijing.aliyuncs.com/docs/Oneflow-Inc/oneflow/pr/10532/

github-actions · 2024-05-30T07:20:56Z

View latest API docs preview at: https://oneflow-staging.oss-cn-beijing.aliyuncs.com/docs/Oneflow-Inc/oneflow/pr/10532/

github-actions · 2024-05-30T08:04:26Z

Speed stats:

GPU Name: NVIDIA GeForce RTX 3080 Ti 

❌ OneFlow resnet50 time: 43.3ms (= 4328.7ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 57.3ms (= 5731.2ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.32 (= 57.3ms / 43.3ms)

OneFlow resnet50 time: 26.5ms (= 2651.8ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 37.6ms (= 3755.2ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.42 (= 37.6ms / 26.5ms)

OneFlow resnet50 time: 18.7ms (= 3735.0ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 34.7ms (= 6949.2ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.86 (= 34.7ms / 18.7ms)

OneFlow resnet50 time: 17.4ms (= 3474.1ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 30.3ms (= 6050.0ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.74 (= 30.3ms / 17.4ms)

OneFlow resnet50 time: 16.7ms (= 3334.1ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 32.2ms (= 6437.5ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.93 (= 32.2ms / 16.7ms)

OneFlow swin dataloader time: 0.200s (= 39.979s / 200, num_workers=1)
PyTorch swin dataloader time: 0.128s (= 25.612s / 200, num_workers=1)
Relative speed: 0.641 (= 0.128s / 0.200s)

OneFlow swin dataloader time: 0.056s (= 11.218s / 200, num_workers=4)
PyTorch swin dataloader time: 0.032s (= 6.494s / 200, num_workers=4)
Relative speed: 0.579 (= 0.032s / 0.056s)

OneFlow swin dataloader time: 0.031s (= 6.238s / 200, num_workers=8)
PyTorch swin dataloader time: 0.017s (= 3.338s / 200, num_workers=8)
Relative speed: 0.535 (= 0.017s / 0.031s)

❌ OneFlow resnet50 time: 49.3ms (= 4928.4ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 63.9ms (= 6389.1ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.30 (= 63.9ms / 49.3ms)

OneFlow resnet50 time: 37.3ms (= 3734.7ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 46.2ms (= 4622.2ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.24 (= 46.2ms / 37.3ms)

OneFlow resnet50 time: 27.7ms (= 5539.0ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 39.7ms (= 7941.4ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.43 (= 39.7ms / 27.7ms)

OneFlow resnet50 time: 25.2ms (= 5030.7ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 38.5ms (= 7699.5ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.53 (= 38.5ms / 25.2ms)

OneFlow resnet50 time: 25.2ms (= 5034.7ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 36.2ms (= 7239.3ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.44 (= 36.2ms / 25.2ms)

mosout · 2024-05-30T08:15:20Z

oneflow/core/functional/impl/common.h

                                  const std::shared_ptr<Tensor>& x_cast);
 Maybe<void> CheckInplaceShapeCanExpandTo(const Shape& shape, const Shape& expand_shape);
+
+inline Maybe<void> CheckSizeNonNegative(const Shape& shape) {


这个名字改成CheckShapeNonNegative吧

mosout · 2024-05-30T08:16:26Z

python/oneflow/nn/modules/pooling.py

        self.padding = padding

    def forward(self, x, indices, output_size=None):
+        kernel_size = _single(self.kernel_size)


重复逻辑封装成函数

github-actions · 2024-05-31T06:18:55Z

Code got formatted by CI. Please request CI again if you still want to have this PR merged. If the PR is from a forked repo, please download the patch files from the GitHub Actions web page and apply them locally.

github-actions · 2024-05-31T06:50:57Z

View latest API docs preview at: https://oneflow-staging.oss-cn-beijing.aliyuncs.com/docs/Oneflow-Inc/oneflow/pr/10532/

github-actions · 2024-05-31T07:31:26Z

Speed stats:

GPU Name: NVIDIA GeForce RTX 3080 Ti 

❌ OneFlow resnet50 time: 43.3ms (= 4327.3ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 57.2ms (= 5722.0ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.32 (= 57.2ms / 43.3ms)

OneFlow resnet50 time: 26.1ms (= 2612.9ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 38.0ms (= 3795.8ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.45 (= 38.0ms / 26.1ms)

OneFlow resnet50 time: 18.3ms (= 3658.2ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 35.1ms (= 7019.6ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.92 (= 35.1ms / 18.3ms)

OneFlow resnet50 time: 17.2ms (= 3447.0ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 30.5ms (= 6092.3ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.77 (= 30.5ms / 17.2ms)

OneFlow resnet50 time: 17.0ms (= 3396.4ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 29.6ms (= 5910.8ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.74 (= 29.6ms / 17.0ms)

OneFlow swin dataloader time: 0.202s (= 40.388s / 200, num_workers=1)
PyTorch swin dataloader time: 0.130s (= 25.952s / 200, num_workers=1)
Relative speed: 0.643 (= 0.130s / 0.202s)

OneFlow swin dataloader time: 0.054s (= 10.861s / 200, num_workers=4)
PyTorch swin dataloader time: 0.032s (= 6.453s / 200, num_workers=4)
Relative speed: 0.594 (= 0.032s / 0.054s)

OneFlow swin dataloader time: 0.030s (= 6.054s / 200, num_workers=8)
PyTorch swin dataloader time: 0.016s (= 3.269s / 200, num_workers=8)
Relative speed: 0.540 (= 0.016s / 0.030s)

❌ OneFlow resnet50 time: 49.3ms (= 4927.4ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 64.5ms (= 6452.8ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.31 (= 64.5ms / 49.3ms)

OneFlow resnet50 time: 36.5ms (= 3648.7ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 46.8ms (= 4675.5ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.28 (= 46.8ms / 36.5ms)

OneFlow resnet50 time: 27.8ms (= 5554.1ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 42.0ms (= 8407.7ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.51 (= 42.0ms / 27.8ms)

OneFlow resnet50 time: 25.2ms (= 5042.9ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 38.7ms (= 7734.2ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.53 (= 38.7ms / 25.2ms)

OneFlow resnet50 time: 25.3ms (= 5051.3ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 36.3ms (= 7255.2ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.44 (= 36.3ms / 25.3ms)

github-actions · 2024-06-02T16:11:10Z

View latest API docs preview at: https://oneflow-staging.oss-cn-beijing.aliyuncs.com/docs/Oneflow-Inc/oneflow/pr/10532/

github-actions · 2024-06-02T16:52:11Z

Speed stats:

GPU Name: NVIDIA GeForce RTX 3080 Ti 

❌ OneFlow resnet50 time: 44.0ms (= 4399.0ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 57.3ms (= 5729.8ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.30 (= 57.3ms / 44.0ms)

OneFlow resnet50 time: 26.6ms (= 2659.2ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 37.6ms (= 3758.2ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.41 (= 37.6ms / 26.6ms)

OneFlow resnet50 time: 18.5ms (= 3703.2ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 34.8ms (= 6956.0ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.88 (= 34.8ms / 18.5ms)

OneFlow resnet50 time: 17.2ms (= 3440.9ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 32.1ms (= 6414.7ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.86 (= 32.1ms / 17.2ms)

OneFlow resnet50 time: 17.0ms (= 3402.2ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 29.8ms (= 5950.3ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.75 (= 29.8ms / 17.0ms)

OneFlow swin dataloader time: 0.200s (= 39.978s / 200, num_workers=1)
PyTorch swin dataloader time: 0.128s (= 25.624s / 200, num_workers=1)
Relative speed: 0.641 (= 0.128s / 0.200s)

OneFlow swin dataloader time: 0.055s (= 11.015s / 200, num_workers=4)
PyTorch swin dataloader time: 0.032s (= 6.459s / 200, num_workers=4)
Relative speed: 0.586 (= 0.032s / 0.055s)

OneFlow swin dataloader time: 0.031s (= 6.118s / 200, num_workers=8)
PyTorch swin dataloader time: 0.017s (= 3.378s / 200, num_workers=8)
Relative speed: 0.552 (= 0.017s / 0.031s)

❌ OneFlow resnet50 time: 49.5ms (= 4952.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 65.4ms (= 6538.7ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.32 (= 65.4ms / 49.5ms)

OneFlow resnet50 time: 37.0ms (= 3703.0ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 47.3ms (= 4729.2ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.28 (= 47.3ms / 37.0ms)

OneFlow resnet50 time: 28.0ms (= 5598.1ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 39.7ms (= 7940.1ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.42 (= 39.7ms / 28.0ms)

OneFlow resnet50 time: 25.3ms (= 5055.0ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 38.7ms (= 7734.2ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.53 (= 38.7ms / 25.3ms)

OneFlow resnet50 time: 24.9ms (= 4989.8ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 36.0ms (= 7198.5ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.44 (= 36.0ms / 24.9ms)

github-actions · 2024-07-10T02:42:13Z

View latest API docs preview at: https://oneflow-staging.oss-cn-beijing.aliyuncs.com/docs/Oneflow-Inc/oneflow/pr/10532/

github-actions · 2024-07-10T03:26:14Z

Speed stats:

GPU Name: NVIDIA GeForce RTX 3080 Ti 

❌ OneFlow resnet50 time: 43.6ms (= 4355.9ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 57.3ms (= 5732.2ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.32 (= 57.3ms / 43.6ms)

OneFlow resnet50 time: 26.1ms (= 2609.7ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 37.4ms (= 3741.4ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.43 (= 37.4ms / 26.1ms)

OneFlow resnet50 time: 18.5ms (= 3709.0ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 35.7ms (= 7149.9ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.93 (= 35.7ms / 18.5ms)

OneFlow resnet50 time: 16.8ms (= 3356.2ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 32.6ms (= 6514.5ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.94 (= 32.6ms / 16.8ms)

OneFlow resnet50 time: 17.2ms (= 3441.1ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 29.0ms (= 5792.1ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.68 (= 29.0ms / 17.2ms)

OneFlow swin dataloader time: 0.201s (= 40.291s / 200, num_workers=1)
PyTorch swin dataloader time: 0.127s (= 25.486s / 200, num_workers=1)
Relative speed: 0.633 (= 0.127s / 0.201s)

OneFlow swin dataloader time: 0.058s (= 11.585s / 200, num_workers=4)
PyTorch swin dataloader time: 0.032s (= 6.479s / 200, num_workers=4)
Relative speed: 0.559 (= 0.032s / 0.058s)

OneFlow swin dataloader time: 0.030s (= 6.085s / 200, num_workers=8)
PyTorch swin dataloader time: 0.017s (= 3.348s / 200, num_workers=8)
Relative speed: 0.550 (= 0.017s / 0.030s)

❌ OneFlow resnet50 time: 49.2ms (= 4920.8ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 64.7ms (= 6465.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.31 (= 64.7ms / 49.2ms)

OneFlow resnet50 time: 37.3ms (= 3729.9ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 46.1ms (= 4614.4ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.24 (= 46.1ms / 37.3ms)

OneFlow resnet50 time: 27.7ms (= 5530.7ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 40.8ms (= 8156.8ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.47 (= 40.8ms / 27.7ms)

OneFlow resnet50 time: 25.4ms (= 5071.2ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 39.2ms (= 7845.1ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.55 (= 39.2ms / 25.4ms)

OneFlow resnet50 time: 24.9ms (= 4972.8ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 35.6ms (= 7115.2ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.43 (= 35.6ms / 24.9ms)

Dmovic added 4 commits May 23, 2024 05:53

add adaptive avg pool input check

3614247

add max unpool input check

148146f

add adaptive max pool input check

e8784e3

fix max pool2d

64deba5

Dmovic requested a review from mosout May 24, 2024 02:43

auto format by CI

2a24cd7

Dmovic and others added 5 commits May 27, 2024 11:11

add Tensor shape check

ef5005c

auto format by CI

33766ca

add rand shape check

a1daa40

fix non negative

2481e92

add zeros ones shape check

65dceb2

Dmovic requested a review from hjchen2 as a code owner May 28, 2024 08:15

auto format by CI

ad0986a

Dmovic requested review from hjchen2 and removed request for hjchen2 May 28, 2024 08:16

auto format by CI

170a197

Dmovic added 2 commits May 29, 2024 08:04

add check non negative function

656879e

Merge branch 'support_input_check' of https://github.com/Oneflow-Inc/…

0a78cdc

…oneflow into support_input_check

Oneflow-Inc deleted a comment from github-actions bot May 29, 2024

update check non negative

b0881a8

Dmovic force-pushed the support_input_check branch from 414a637 to b0881a8 Compare May 29, 2024 08:58

Dmovic added 4 commits May 29, 2024 09:01

update check negative

0bc50c9

inline check non negative

4548aac

add empty shape check

314a3cc

update adaptive none

df929ec

auto format by CI

3438b2f

add randn shape check

4ce4bab

mosout reviewed May 30, 2024

View reviewed changes

Dmovic and others added 2 commits May 31, 2024 06:16

update check function

4dc75dc

auto format by CI

f0e5e24

Dmovic added the enhancement label May 31, 2024

Dmovic removed the enhancement label May 31, 2024

Dmovic requested a review from oneflow-ci-bot June 2, 2024 15:37

Dmovic added system enhancement labels Jun 2, 2024

Dmovic removed the request for review from oneflow-ci-bot June 3, 2024 04:19

Merge branch 'master' into support_input_check

0d56eba

Dmovic requested review from oneflow-ci-bot and removed request for oneflow-ci-bot July 10, 2024 02:08

Support input check for pool operator #10532

Are you sure you want to change the base?

Support input check for pool operator #10532

Uh oh!

Conversation

Dmovic commented May 24, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

CLAassistant commented May 24, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions bot commented May 24, 2024

Uh oh!

github-actions bot commented May 28, 2024

Uh oh!

github-actions bot commented May 28, 2024

Uh oh!

github-actions bot commented May 29, 2024

Uh oh!

github-actions bot commented May 30, 2024

Uh oh!

github-actions bot commented May 30, 2024

Uh oh!

github-actions bot commented May 30, 2024

Uh oh!

github-actions bot commented May 30, 2024

Uh oh!

mosout May 30, 2024

Choose a reason for hiding this comment

Uh oh!

mosout May 30, 2024

Choose a reason for hiding this comment

Uh oh!

github-actions bot commented May 31, 2024

Uh oh!

github-actions bot commented May 31, 2024

Uh oh!

github-actions bot commented May 31, 2024

Uh oh!

github-actions bot commented Jun 2, 2024

Uh oh!

github-actions bot commented Jun 2, 2024

Uh oh!

github-actions bot commented Jul 10, 2024

Uh oh!

github-actions bot commented Jul 10, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Dmovic commented May 24, 2024 •

edited

Loading

CLAassistant commented May 24, 2024 •

edited

Loading