oneflow
oneflow copied to clipboard
Published
20 hours ago •
Oneflow-Inc
Oneflow-Inc
add mse_loss and ls_loss interface
添加了一下nn.functional.mse_loss()和nn.functional.l1_loss()接口和document https://github.com/Oneflow-Inc/OneTeam/issues/1207#issuecomment-1073432125
View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/8884/
![github-actions[bot] avatar](https://avatars.githubusercontent.com/in/15368?v=4 "Published by a pub.dev verified publisher"){kind=small}
Speed stats:
GPU Name: GeForce GTX 1080
✔️ OneFlow resnet50 time: 128.4ms (= 12842.1ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 140.5ms (= 14050.0ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.09 (= 140.5ms / 128.4ms)
OneFlow resnet50 time: 75.6ms (= 7560.4ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 82.7ms (= 8265.2ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.09 (= 82.7ms / 75.6ms)
OneFlow resnet50 time: 48.4ms (= 9681.7ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 52.9ms (= 10587.8ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.09 (= 52.9ms / 48.4ms)
OneFlow resnet50 time: 35.9ms (= 7188.6ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 45.4ms (= 9088.2ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.26 (= 45.4ms / 35.9ms)
OneFlow resnet50 time: 28.1ms (= 5618.5ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 36.3ms (= 7253.4ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.29 (= 36.3ms / 28.1ms)
OneFlow swin dataloader time: 0.275s (= 55.094s / 200, num_workers=1)
PyTorch swin dataloader time: 0.150s (= 29.943s / 200, num_workers=1)
Relative speed: 0.543 (= 0.150s / 0.275s)
OneFlow swin dataloader time: 0.071s (= 14.283s / 200, num_workers=4)
PyTorch swin dataloader time: 0.042s (= 8.455s / 200, num_workers=4)
Relative speed: 0.592 (= 0.042s / 0.071s)
OneFlow swin dataloader time: 0.041s (= 8.253s / 200, num_workers=8)
PyTorch swin dataloader time: 0.022s (= 4.467s / 200, num_workers=8)
Relative speed: 0.541 (= 0.022s / 0.041s)
❌ OneFlow resnet50 time: 136.7ms (= 13665.7ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 161.9ms (= 16185.7ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.18 (= 161.9ms / 136.7ms)
OneFlow resnet50 time: 84.2ms (= 8420.2ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 113.2ms (= 11318.0ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.34 (= 113.2ms / 84.2ms)
OneFlow resnet50 time: 57.4ms (= 11475.6ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 77.7ms (= 15535.5ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.35 (= 77.7ms / 57.4ms)
OneFlow resnet50 time: 45.4ms (= 9072.7ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 71.2ms (= 14246.5ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.57 (= 71.2ms / 45.4ms)
OneFlow resnet50 time: 38.9ms (= 7789.1ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 69.5ms (= 13898.3ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.78 (= 69.5ms / 38.9ms)
可以按照 oneflow/python/oneflow/test/modules/test_loss.py 这里的代码，加一下 nn.functional.l1_loss / mse_loss 的单测；文档中可以加一下参数列表，不然渲染出来文档oneflow.nn.functional.mse_loss后面只跟了一个空括号。其他没啥问题了
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.
View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/8884/
Speed stats:
GPU Name: GeForce GTX 1080
✔️ OneFlow resnet50 time: 128.4ms (= 12843.6ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 146.7ms (= 14665.3ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.14 (= 146.7ms / 128.4ms)
OneFlow resnet50 time: 75.4ms (= 7538.4ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 83.1ms (= 8307.3ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.10 (= 83.1ms / 75.4ms)
OneFlow resnet50 time: 48.6ms (= 9723.3ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 58.1ms (= 11628.4ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.20 (= 58.1ms / 48.6ms)
OneFlow resnet50 time: 36.2ms (= 7232.6ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 42.8ms (= 8550.6ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.18 (= 42.8ms / 36.2ms)
OneFlow resnet50 time: 28.4ms (= 5674.9ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 41.3ms (= 8253.0ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.45 (= 41.3ms / 28.4ms)
OneFlow swin dataloader time: 0.271s (= 54.200s / 200, num_workers=1)
PyTorch swin dataloader time: 0.151s (= 30.171s / 200, num_workers=1)
Relative speed: 0.557 (= 0.151s / 0.271s)
OneFlow swin dataloader time: 0.110s (= 22.050s / 200, num_workers=4)
PyTorch swin dataloader time: 0.042s (= 8.331s / 200, num_workers=4)
Relative speed: 0.378 (= 0.042s / 0.110s)
OneFlow swin dataloader time: 0.040s (= 8.007s / 200, num_workers=8)
PyTorch swin dataloader time: 0.024s (= 4.703s / 200, num_workers=8)
Relative speed: 0.587 (= 0.024s / 0.040s)
❌ OneFlow resnet50 time: 136.9ms (= 13686.1ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 160.3ms (= 16027.3ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.17 (= 160.3ms / 136.9ms)
OneFlow resnet50 time: 84.9ms (= 8485.9ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 111.4ms (= 11137.0ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.31 (= 111.4ms / 84.9ms)
OneFlow resnet50 time: 58.4ms (= 11680.6ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 88.7ms (= 17748.7ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.52 (= 88.7ms / 58.4ms)
OneFlow resnet50 time: 45.4ms (= 9086.4ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 68.6ms (= 13726.2ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.51 (= 68.6ms / 45.4ms)
OneFlow resnet50 time: 39.2ms (= 7830.5ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 68.2ms (= 13635.8ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.74 (= 68.2ms / 39.2ms)
Speed stats:
GPU Name: GeForce GTX 1080
✔️ OneFlow resnet50 time: 128.2ms (= 12822.4ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 148.9ms (= 14888.7ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.16 (= 148.9ms / 128.2ms)
OneFlow resnet50 time: 75.3ms (= 7528.8ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 84.5ms (= 8449.4ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.12 (= 84.5ms / 75.3ms)
OneFlow resnet50 time: 48.4ms (= 9671.0ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 57.4ms (= 11475.2ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.19 (= 57.4ms / 48.4ms)
OneFlow resnet50 time: 36.1ms (= 7214.3ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 47.7ms (= 9543.6ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.32 (= 47.7ms / 36.1ms)
OneFlow resnet50 time: 28.2ms (= 5640.2ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 46.8ms (= 9366.5ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.66 (= 46.8ms / 28.2ms)
OneFlow swin dataloader time: 0.265s (= 52.994s / 200, num_workers=1)
PyTorch swin dataloader time: 0.151s (= 30.137s / 200, num_workers=1)
Relative speed: 0.569 (= 0.151s / 0.265s)
OneFlow swin dataloader time: 0.072s (= 14.428s / 200, num_workers=4)
PyTorch swin dataloader time: 0.040s (= 8.025s / 200, num_workers=4)
Relative speed: 0.556 (= 0.040s / 0.072s)
OneFlow swin dataloader time: 0.040s (= 7.919s / 200, num_workers=8)
PyTorch swin dataloader time: 0.022s (= 4.370s / 200, num_workers=8)
Relative speed: 0.552 (= 0.022s / 0.040s)
❌ OneFlow resnet50 time: 136.7ms (= 13669.0ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 161.0ms (= 16096.6ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.18 (= 161.0ms / 136.7ms)
OneFlow resnet50 time: 85.0ms (= 8500.2ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 103.2ms (= 10315.1ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.21 (= 103.2ms / 85.0ms)
OneFlow resnet50 time: 57.7ms (= 11546.8ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 85.1ms (= 17012.9ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.47 (= 85.1ms / 57.7ms)
OneFlow resnet50 time: 45.0ms (= 9002.9ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 74.5ms (= 14898.1ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.65 (= 74.5ms / 45.0ms)
OneFlow resnet50 time: 38.9ms (= 7780.5ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 67.0ms (= 13400.5ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.72 (= 67.0ms / 38.9ms)
View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/8884/
View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/8884/
Speed stats:
GPU Name: GeForce GTX 1080
✔️ OneFlow resnet50 time: 128.4ms (= 12840.4ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 149.3ms (= 14928.2ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.16 (= 149.3ms / 128.4ms)
OneFlow resnet50 time: 75.3ms (= 7531.6ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 87.2ms (= 8722.1ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.16 (= 87.2ms / 75.3ms)
OneFlow resnet50 time: 48.5ms (= 9701.9ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 59.1ms (= 11824.9ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.22 (= 59.1ms / 48.5ms)
OneFlow resnet50 time: 36.2ms (= 7232.6ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 41.4ms (= 8280.5ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.14 (= 41.4ms / 36.2ms)
OneFlow resnet50 time: 28.1ms (= 5629.4ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 40.1ms (= 8020.4ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.42 (= 40.1ms / 28.1ms)
OneFlow swin dataloader time: 0.267s (= 53.314s / 200, num_workers=1)
PyTorch swin dataloader time: 0.151s (= 30.280s / 200, num_workers=1)
Relative speed: 0.568 (= 0.151s / 0.267s)
OneFlow swin dataloader time: 0.070s (= 14.021s / 200, num_workers=4)
PyTorch swin dataloader time: 0.042s (= 8.361s / 200, num_workers=4)
Relative speed: 0.596 (= 0.042s / 0.070s)
OneFlow swin dataloader time: 0.040s (= 7.991s / 200, num_workers=8)
PyTorch swin dataloader time: 0.022s (= 4.480s / 200, num_workers=8)
Relative speed: 0.561 (= 0.022s / 0.040s)
❌ OneFlow resnet50 time: 136.7ms (= 13672.2ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 160.5ms (= 16048.0ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.17 (= 160.5ms / 136.7ms)
OneFlow resnet50 time: 84.8ms (= 8483.9ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 112.6ms (= 11256.2ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.33 (= 112.6ms / 84.8ms)
OneFlow resnet50 time: 58.4ms (= 11682.2ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 78.8ms (= 15760.4ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.35 (= 78.8ms / 58.4ms)
OneFlow resnet50 time: 45.4ms (= 9074.6ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 69.4ms (= 13876.9ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.53 (= 69.4ms / 45.4ms)
OneFlow resnet50 time: 38.9ms (= 7770.8ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 67.6ms (= 13523.9ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.74 (= 67.6ms / 38.9ms)