PR #6838 Sync from master

Sync from master #6838

SherlockNoMad merged 38 commits into thiagofc/ortmodule-api from mzs/ortmodule-api-sync-from-master-210226

Bumping up version to 1.7 (#6736)

ea3aee4d

Avoid removing constant weight that is graph output (#6735)

b1a12b49

Upgrade TensorRT to v7.2.2 (#6452)

53eb948f

Update docs MCR image publishing policy change (#6743)

3722dd26

Entropy method for calibration-based quantization (#6619)

67c478ed

. (#6751)

79b966b0

fix link to samples in nodejs README (#6746)

a25c8e52

Add direct link to build instructions on readme (#6729)

20d6613e

Removed BUILD.md from master as source now lives in gh-pages (#6709)

c1700619

Fix iOS/macOS build warning for inconsistent symbol visibility settin…

47519623

Fix longformer parity and perf regression (#6760)

3bda7f4d

Make it easier to fold MakeString usages to reduce binary size. (#6754)

9f7dffb1

Add robust dependency check for Python package (#6436)

c91f3142

Support specifying globally allowed types from build script (#6677)

ee35be01

[CoreML EP] Add options to enable CoreML EP only on hardware with App…

f5313cc4

add --enable_cuda_line_info to build.py (#6773)

8a6f6bc3

update benchmark_longformer for default test suite (#6772)

8703e2c7

Enable CI to cover globally allowed types (#6778)

5db0c9c6

Update docs/ONNX_Runtime_for_Mobile_Platforms.md with info about op t…

09a5d6a9

Update DirectML 1.4.1 to 1.4.2 for ORT 1.7 (#6780)

b8b41e37

Support building python bindings when building unit tests is manually…

7aa69a1f

[Running CI now] Remove duplicate tests to speed up CI (#6768)

c02ec38f

Adding fp16 support for Einsum Cuda kernel (#6775)

47c8e9ad

[Java] Adds extra providers (#6770)

5a473216

Allow 3D ConvTranspose in CUDA execution provider (#6794)

71a70ecf

Update transformers benchmark for transformers 4.3.* and ORT 1.7 (#6796)

f4acdb2e

Enable more unit tests for ROCM EP (#6776)

40fa40f3

Fix regression in constant folding optimizer (#6795)

ab1713f5

make ci pipeline also run batch and convergence test (#6798)

8a148e44

[OpenVINO-EP] Find package changes (#6801)

7465673e

fix issues caused by quantize/calibrate changes (#6802)

d5f292ab

Setup perf in docker and add features (#6582)

db05d53b

Make keepdims to its default value when adding ReduceMin/ReduceMax fo…

9b3171e9

Fixed issue in python cmake to update wheel package (#6384)

34261087

Improvements to quantizer: Removed unused qType field, add reshape op…

d5175795

update (#6827)

b4b87ac7

Do not create compute stream when external CUDA allocator is used. (#…

cb8d8464

Merge branch 'master' of https://github.com/microsoft/onnxruntime int…

ca48310d