strisunshinewentingwang/pytorch

Pull Requests Commits

use ufunc_defs.bzl in BUILD.bazel

Michael Andreas Dagitses committed 2 years ago

c6c207f6

Back out "[const_fold] Set requires_grad based on the folded tensor; add device_for_folding option" (#79655)

singlaiiit committed 2 years ago

24c2aff1

[quant] Add quantized levels visualization

asl3 committed 2 years ago

afc037ae

[tensorboard] update assertion error for `scalar()` and fix docs (#76859)

edward-io committed 2 years ago

0088172e

Remove -Wno-unused-but-set-variable for clang 13.0.0 (#79666)

kit1980 committed 2 years ago

f1fb575b

Reland PySymInt (#79617)

Krovatkin committed 2 years ago

8ef6356f

Use amax instead of max for softmax decomps (#79667)

jansel committed 2 years ago

c8fb02b4

[Profiler] Prepare for change to unique_ptr in Kineto

Taylor Robie committed 2 years ago

ef1e5d0e

[CI] Disable Buck Build Test (#79675)

zengk95 committed 2 years ago

c348980b

[ci] don't upload sccache artifacts with the wrong extension

suo committed 2 years ago

ac0d540d

[ONNX] Improve docstrings and typing annotations (#78231)

justinchuby committed 2 years ago

4b817f58

[FSDP] Extend ignored modules test to not pass to root

awgu committed 2 years ago

9681c75f

use exe/exepath in our genrules

Michael Andreas Dagitses committed 2 years ago

e21c0ac9

fix generate-code caching by indicating that the binary is an executable

Michael Andreas Dagitses committed 2 years ago

86606fbe

[FSDP] Fix exec order validation for diff ignored modules across ranks

awgu committed 2 years ago

18fcd482

Support percentages in random_split (#78877)

bushshrub committed 2 years ago

3064982f

fix at::from_blob_quantized_per_tensor_affine strides calculation (#79314)

yzyz committed 2 years ago

2f437847

Revert "automate hash pinning (#79470)"

pytorchmergebot committed 2 years ago

44e5593b

Make l1_loss composite

lezcano committed 2 years ago

61a5c779

[JIT] Change SchemaCheckTensor into SchemaCheckMode and fix global variable issues

goldenxuett committed 2 years ago

42fd58ea

[ONNX] Fix arg type in `_set_training_mode` (#78583)

justinchuby committed 2 years ago

c8b9b626

[quant] Add param calcs and tests for APoT observers

asl3 committed 2 years ago

81f27700

add a new alias key for functional to view op decompositions

bdhirsh committed 2 years ago

adf80606

[quant][core][improvement][feature] Enabled support for quantized fill of nhwc tensors

dzdang committed 2 years ago

6114b0f9

Update ideep for ideep conv changes -v2 (#79258)

yanbing-j committed 2 years ago

89bae066

update tensor impl support_as_strided for nested tensors (#79616)

drisspg committed 2 years ago

55e3d7a4

Create decoder class (#79438)

erichan1 committed 2 years ago

0ac2289a

Refactor out QKV in projection (#79437)

erichan1 committed 2 years ago

110c7306

[FSDP] Add forward prefetching option in FSDP API (#78841)

LinjianMa committed 2 years ago

70446c25

Made improvements to isGreen (#79619)

swang392 committed 2 years ago

26e1afe9

Newer Older