…2db60a

Summary:
Previous import was 96c58ceeacf0f2b73d752e413e4fd78787a12da3

Included changes:
- **[3b0d3bb](onnx/onnx@3b0d3bb)**: remove global variable in header file (pytorch#1850) <Lu Fang>
- **[1cca873](onnx/onnx@1cca873)**: bump the version for drop out - fix the issue that the version was not bumped when changing its type constraint declaration. (pytorch#1848) <Ke Zhang>
- **[1ec81bc](onnx/onnx@1ec81bc)**: Change TopK operator to allow dynamic 'k' (pytorch#1829) <Hariharan Seshadri>
- **[a89a4a1](onnx/onnx@a89a4a1)**: Remove exp op: Affine, ImageScaler,ParametricSoftplus, Crop. (pytorch#1832) <Ke Zhang>

Differential Revision: D14395214

fbshipit-source-id: a43bb11171ef7a9873239f57ed6c61b051133afe

…86aef9 (pytorch#18248)

Summary:
Pull Request resolved: pytorch#18248

Previous import was 96c58ceeacf0f2b73d752e413e4fd78787a12da3

Included changes:
- **[f6f80657](onnx/onnx@f6f80657)**: Skip the schema check on ops in non-standard domain (pytorch#1876) <Lu Fang>
- **[8c8be722](onnx/onnx@8c8be722)**: Introduce Function Body Helper  (pytorch#1868) <Sherlock>
- **[b605eafb](onnx/onnx@b605eafb)**: Support down sampling for Upsample with scales < 1. (pytorch#1773) <Ke Zhang>
- **[47f7aa71](onnx/onnx@47f7aa71)**: Remove scaledtanh (pytorch#1866) <Ashwini Khade>
- **[4dfc56de](onnx/onnx@4dfc56de)**: Add Ceil support for Max and Average Pooling (pytorch#1860) <Lara Haidar>
- **[552a8efc](onnx/onnx@552a8efc)**: Add testcase generator for functions (pytorch#1862) <Raymond Yang>
- **[fdb978a5](onnx/onnx@fdb978a5)**: Promote Thresholded Relu Op (pytorch#1856) <Ashwini Khade>
- **[ce332628](onnx/onnx@ce332628)**: Update Slice with dynamic input & optional input steps (pytorch#1836) <Bowen Bao>
- **[3a9a8787](onnx/onnx@3a9a8787)**: Merge function into opschema (pytorch#1834) <Raymond Yang>
- **[3dbf8fe9](onnx/onnx@3dbf8fe9)**: Handle string comparision represented as np.objects (pytorch#1851) <Dmitri Smirnov>
- **[3b0d3bb2](onnx/onnx@3b0d3bb2)**: remove global variable in header file (pytorch#1850) <Lu Fang>
- **[1cca8733](onnx/onnx@1cca8733)**: bump the version for drop out - fix the issue that the version was not bumped when changing its type constraint declaration. (pytorch#1848) <Ke Zhang>
- **[1ec81bc6](onnx/onnx@1ec81bc6)**: Change TopK operator to allow dynamic 'k' (pytorch#1829) <Hariharan Seshadri>
- **[a89a4a16](onnx/onnx@a89a4a16)**: Remove exp op: Affine, ImageScaler,ParametricSoftplus, Crop. (pytorch#1832) <Ke Zhang>

Differential Revision: D14549289

fbshipit-source-id: 1222721e9766d30d559ad7a5fba6ba0a6afd6344

…e0ea6c (pytorch#18285)

Summary:
Pull Request resolved: pytorch#18285

Previous import was 96c58ceeacf0f2b73d752e413e4fd78787a12da3

Included changes:
- **[c05f2ae4](onnx/onnx@c05f2ae4)**: update both core and ml docs (pytorch#1879) <Lu Fang>
- **[f895279b](onnx/onnx@f895279b)**: fix the problems introduced in previous PRs in operator registration (pytorch#1878) <Lu Fang>
- **[f6f80657](onnx/onnx@f6f80657)**: Skip the schema check on ops in non-standard domain (pytorch#1876) <Lu Fang>
- **[8c8be722](onnx/onnx@8c8be722)**: Introduce Function Body Helper  (pytorch#1868) <Sherlock>
- **[b605eafb](onnx/onnx@b605eafb)**: Support down sampling for Upsample with scales < 1. (pytorch#1773) <Ke Zhang>
- **[47f7aa71](onnx/onnx@47f7aa71)**: Remove scaledtanh (pytorch#1866) <Ashwini Khade>
- **[4dfc56de](onnx/onnx@4dfc56de)**: Add Ceil support for Max and Average Pooling (pytorch#1860) <Lara Haidar>
- **[552a8efc](onnx/onnx@552a8efc)**: Add testcase generator for functions (pytorch#1862) <Raymond Yang>
- **[fdb978a5](onnx/onnx@fdb978a5)**: Promote Thresholded Relu Op (pytorch#1856) <Ashwini Khade>
- **[ce332628](onnx/onnx@ce332628)**: Update Slice with dynamic input & optional input steps (pytorch#1836) <Bowen Bao>
- **[3a9a8787](onnx/onnx@3a9a8787)**: Merge function into opschema (pytorch#1834) <Raymond Yang>
- **[3dbf8fe9](onnx/onnx@3dbf8fe9)**: Handle string comparision represented as np.objects (pytorch#1851) <Dmitri Smirnov>
- **[3b0d3bb2](onnx/onnx@3b0d3bb2)**: remove global variable in header file (pytorch#1850) <Lu Fang>
- **[1cca8733](onnx/onnx@1cca8733)**: bump the version for drop out - fix the issue that the version was not bumped when changing its type constraint declaration. (pytorch#1848) <Ke Zhang>
- **[1ec81bc6](onnx/onnx@1ec81bc6)**: Change TopK operator to allow dynamic 'k' (pytorch#1829) <Hariharan Seshadri>
- **[a89a4a16](onnx/onnx@a89a4a16)**: Remove exp op: Affine, ImageScaler,ParametricSoftplus, Crop. (pytorch#1832) <Ke Zhang>

Differential Revision: D14566202

fbshipit-source-id: 3deb51c17eb9ebd6f6efc331d9110acb3462ece5

…e0ea6c (#18285)

Summary:
Pull Request resolved: #18285

Previous import was 96c58ceeacf0f2b73d752e413e4fd78787a12da3

Included changes:
- **[c05f2ae4](onnx/onnx@c05f2ae4)**: update both core and ml docs (#1879) <Lu Fang>
- **[f895279b](onnx/onnx@f895279b)**: fix the problems introduced in previous PRs in operator registration (#1878) <Lu Fang>
- **[f6f80657](onnx/onnx@f6f80657)**: Skip the schema check on ops in non-standard domain (#1876) <Lu Fang>
- **[8c8be722](onnx/onnx@8c8be722)**: Introduce Function Body Helper  (#1868) <Sherlock>
- **[b605eafb](onnx/onnx@b605eafb)**: Support down sampling for Upsample with scales < 1. (#1773) <Ke Zhang>
- **[47f7aa71](onnx/onnx@47f7aa71)**: Remove scaledtanh (#1866) <Ashwini Khade>
- **[4dfc56de](onnx/onnx@4dfc56de)**: Add Ceil support for Max and Average Pooling (#1860) <Lara Haidar>
- **[552a8efc](onnx/onnx@552a8efc)**: Add testcase generator for functions (#1862) <Raymond Yang>
- **[fdb978a5](onnx/onnx@fdb978a5)**: Promote Thresholded Relu Op (#1856) <Ashwini Khade>
- **[ce332628](onnx/onnx@ce332628)**: Update Slice with dynamic input & optional input steps (#1836) <Bowen Bao>
- **[3a9a8787](onnx/onnx@3a9a8787)**: Merge function into opschema (#1834) <Raymond Yang>
- **[3dbf8fe9](onnx/onnx@3dbf8fe9)**: Handle string comparision represented as np.objects (#1851) <Dmitri Smirnov>
- **[3b0d3bb2](onnx/onnx@3b0d3bb2)**: remove global variable in header file (#1850) <Lu Fang>
- **[1cca8733](onnx/onnx@1cca8733)**: bump the version for drop out - fix the issue that the version was not bumped when changing its type constraint declaration. (#1848) <Ke Zhang>
- **[1ec81bc6](onnx/onnx@1ec81bc6)**: Change TopK operator to allow dynamic 'k' (#1829) <Hariharan Seshadri>
- **[a89a4a16](onnx/onnx@a89a4a16)**: Remove exp op: Affine, ImageScaler,ParametricSoftplus, Crop. (#1832) <Ke Zhang>

Reviewed By: yinghai

Differential Revision: D14566202

fbshipit-source-id: b1e5912ae6887e2865fc628363071e2b9938dfa4

…h#1832)

This is backported from upstream PR pytorch#140172, pytorch#137443 and pytorch#139432.

Original commit message of pytorch#140172:

Notable new features for SDPA operators on AMD systems from AOTriton
0.8b:

1. Nestedtensor support;
2. MQA/GQA support;
3. Restore Efficient attention support for causal=True and seqlen_q !=
seqlen_k cases;
+ The kernel should use top-left alignment, bottom right alignment will
be added later
4. Move gfx1100 (RX7900/W7800/W7900) out of experimental support status.
However, users are strongly recommended to update to ROCM 6.2.4, notably
for its firmware updates.

Related unit tests are enabled as well.

Notable related changes from AOTriton 0.8b:

1. AOTriton 0.8b moves the GPU kernel out of libaotriton.so to a
separate directory `aotriton.images`;
2. LZMA replaces ZSTD as GPU kernel compression algorithm for better
compression ratio: aotriton0.8b (.so + aotriton.images take 350MB)
compared to aotriton0.7b .so: 800MB
3. The compression cannot be disabled now, and `liblzma` is hard
run-time dependency.
+ Should not be a problem, since `lzma` is part of Python Standard
Library

Pull Request resolved: pytorch#140172
Approved by: https://github.com/jithunnair-amd,
https://github.com/jeffdaily

Fixes #ISSUE_NUMBER

---------

Co-authored-by: Jithun Nair <37884920+jithunnair-amd@users.noreply.github.com>