OSSForks/pytorch
1
0
Fork 0
mirror of https://github.com/zebrajr/pytorch.git synced 2025-12-07 00:21:07 +01:00
Code Issues Actions 38 Packages Projects Releases Wiki Activity
528158af47
pytorch/caffe2
History
Zeliang Chen 38c97fb6f0 [shape inference] add shape inference support 
Summary:
* To make pruning op compatible with shape inference, we introduced a new quantile argument (as in D23463390) to differentiate dynamic/fixed pruning.

* The fixed pruning op has defined output shapes. However, the input shapes are not determined therefore we want to bypass the input shapes checking for two pruning ops, as implemented in this diff.

Test Plan:
buck test caffe2/caffe2/opt:bound_shape_inference_test

```
Started reporting to test run: https://our.intern.facebook.com/intern/testinfra/testrun/844425102187909
    ✓ ListingSuccess: caffe2/caffe2/opt:bound_shape_inference_test - main (1.973)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.FC3D (2.604)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.SparseLengthsSumFused4BitRowwise (2.635)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.FC (2.690)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.Int8QuantizeInferInputBackwards (2.705)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.SparseLengthsSum (2.729)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.Reshape (2.754)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.ConcatMissingInput (2.770)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.ElementwiseOp (2.770)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.Tile (2.785)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.Bucketize (2.789)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.SparseLengthsSumFused8BitRowwise (2.807)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.SparseLengthsSum8BitRowwiseSparse (2.841)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.Split (2.863)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.ConcatInferInputBackwards (2.894)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.ElementwiseInferInputBackwards (2.898)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.Combo0 (2.902)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.LengthsRangeFill (2.964)
    ✓ Pass: caffe2/caffe2/opt:bound_shape_inference_test - BoundShapeInference.Quantization (2.964)
Summary
  Pass: 18
  ListingSuccess: 1
Finished test run: https://our.intern.facebook.com/intern/testinfra/testrun/844425102187909
```

buck test caffe2/caffe2/fb/opt:bound_shape_inference_net_test

```
 Started reporting to test run: https://our.intern.facebook.com/intern/testinfra/testrun/3096224780078093
    ✓ ListingSuccess: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - main (14.092)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.ClipLengths (15.508)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.DPER3IdListFeaturePreProcessing (15.521)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.ClipRanges (16.198)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.RowwisePrune (16.302)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - FbBoundShapeInferencerTest.GatherRanges1 (16.585)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.Combo3 (16.865)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.DPER3IdListFeaturePreProcessingWithCast (16.907)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.GatherRanges2 (16.921)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - FbBoundShapeInferencerTest.LengthsRangeFill (17.157)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.ClipRangesAndGatherRanges (17.277)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.DPER3IdScoreListFeaturePreProcessing (17.274)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.ClipRangesGatherSigridHash (17.554)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.Combo1 (17.645)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.DPER3IdScoreListFeaturePreProcessingDEFAULT (17.887)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.DPER3IdListFeaturePreProcessingDEFAULT (17.929)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.f97293388_0 (19.343)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - FbBoundShapeInferencerTest.GatherRangesToDense1 (19.489)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.DPER3IdScoreListFeaturePreProcessingWithCast (19.887)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.xray_v11 (19.905)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - FbBoundShapeInferencerTest.SigridTransforms (20.080)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.Combo2 (20.086)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.vanillaSparseNN (59.847)
    ✓ Pass: caffe2/caffe2/fb/opt:bound_shape_inference_net_test - BoundShapeInference.gather (97.822)
Summary
  Pass: 23
  ListingSuccess: 1
```

## Workflow testing

===
* non-DI/fixed quantile/user side/non-self-binning
f224250571

*  non-DI/fixed quantile/user+ad side/non-self-binning
f224250610

* DI/fixed quantile/user side/self-binning
f224250637

* DI/fixed quantile/user+ad side/self-binning
f224250662

*  non-DI/dynamic quantile/user+ad side/non-self-binning
f224250705

* DI/dynamic quantile/user+ad side/self-binning
f224250760

Reviewed By: ChunliF

Differential Revision: D23647390

fbshipit-source-id: 3ec1c0eaea53bd4d5eda4a0436577216f7fa8ead
2020-10-15 00:46:06 -07:00
..
contrib [FakeLowP] T76913842 Make AddFakeFp16 take int inputs (#45992) 2020-10-07 17:43:00 -07:00
core [caffe2] add PlanExecutorTest ErrorPlanWithCancellableStuckNet (#46110) 2020-10-12 12:00:15 -07:00
cuda_rtc
db
distributed [Distributed] DeleteKey API for c10d TCP Store (#45401) 2020-09-28 15:30:39 -07:00
experiments Remove __future__ imports for legacy Python2 supports (#45033) 2020-09-23 17:57:02 -07:00
ideep Ensure IDEEP transpose operator works correctly 2020-08-11 12:58:31 -07:00
image
mobile Make msg() and msg_with_backtrace() private (#37094) 2020-05-04 11:54:34 -07:00
mpi [codemod][caffe2] Run clang-format - 5/7 2020-06-30 15:45:11 -07:00
observers [codemod][caffe2] Run clang-format - 5/7 2020-06-30 15:45:11 -07:00
onnx Add python inferface to create OfflineTensor (#42516) 2020-08-04 01:31:34 -07:00
operators [caffe2] Add operator schema for FP16SparseNorm (#46300) 2020-10-13 18:58:23 -07:00
opt [shape inference] add shape inference support 2020-10-15 00:46:06 -07:00
perfkernels [caffe2] Fix preprocessor checks for FMA 2020-10-11 11:48:32 -07:00
predictor [caffe2] Do not run RemoveOpsByType on recurrent networks (#45986) 2020-10-07 14:07:52 -07:00
proto [shape inference] adding a new flag to the struct 2020-10-09 19:29:08 -07:00
python [caffe2] Add 10s deadline for all Caffe2 hypothesis fuzz tests 2020-10-14 06:30:09 -07:00
quantization Adding Histogram Binning Calibration to DSNN and Adding Type Double to Caffe2 ParallelSumOp/SumReluOp 2020-09-28 15:21:31 -07:00
queue [caffe2] adds Cancel to SafeDequeueBlobsOp and SafeEnqueueBlobsOp (#45177) 2020-09-24 14:22:46 -07:00
serialize Enable TF32 support for cuDNN (#40737) 2020-09-01 15:34:24 -07:00
sgd [caffe2] Fix preprocessor checks for FMA 2020-10-11 11:48:32 -07:00
share Update pthreadpool to pthreadpool:029c88620802e1361ccf41d1970bd5b07fd6b7bb. (#40524) 2020-07-09 10:00:36 -07:00
test
transforms Replaced whitelisted with allowed (#41867) 2020-07-23 16:53:51 -07:00
utils [Onnxifi] Don't throw exception when we cannot write out debug files (#45979) 2020-10-08 00:18:24 -07:00
video [ROCm][Caffe2] Enable MIOpen 3D Pooling (#38260) 2020-07-08 17:42:55 -07:00
__init__.py remediation of S205607 2020-07-17 17:19:47 -07:00
.clang-format
c2_aten_srcs.bzl
CMakeLists.txt Enable CUDA Fuser for ROCm (#45965) 2020-10-08 10:41:56 -07:00
README.md Grammar patch 1 (.md) (#41599) 2020-07-20 10:19:08 -07:00
release-notes.md
requirements.txt
VERSION_NUMBER

README.md

Caffe2

Jenkins Build Status

Caffe2 is a lightweight, modular, and scalable deep learning framework. Building on the original Caffe, Caffe2 is designed with expression, speed, and modularity in mind.

Questions and Feedback

Please use GitHub issues (https://github.com/pytorch/pytorch/issues) to ask questions, report bugs, and request new features.

Further Resources on Caffe2.ai