strisunshinewentingwang/pytorch

Pull Requests Commits

*_scatter ops should preserve input stride/storage_offset (#91029)

bdhirsh committed 1 year ago

c47bdd75

buck-related minifier work (#91215)

anijain2305 committed 1 year ago

a3291619

[BE] Reformat ReduceOps (#91221)

malfet committed 1 year ago

d397f414

Fix passing frame to callback (#91170)

albanD committed 1 year ago

c7302075

Revert "use scatter_add for index_add when dim is the most inner dim (#88729)"

pytorchmergebot committed 1 year ago

eadd5572

[CUDA12] Clean up deprecated APIs (#91050)

xw285cornell committed 1 year ago

bacd2ced

Fix test_profiler_seq_nr flakiness (on macos) (#91019)

huydhn committed 1 year ago

e40e4d36

[inductor] CI improvments (#91283)

desertfire committed 1 year ago

07c61685

[dynamo] Write full code of how to enable `output_code` (#91230)

take-cheeze committed 1 year ago

55749b9c

Fix for `mul(compressed, wrapped scalar)` (#91239)

nikitaved committed 1 year ago

4c5928e3

[inductor] New approach for computing triton load/store masks (#91241)

Fabio Rocha committed 1 year ago

bc843682

[MPS] Add `group_norm[fwd+backward]` and `mean_var` (take 2) (#91190)

malfet committed 1 year ago

fd3a7264

[Composable API] Make _StateKey as a str subclass (#91279)

fegin committed 1 year ago

9b42e4ef

Use `@pytorch//` in bazel build files (#89660)

vors committed 1 year ago

39306c1d

[FSDP][optim_state_dict][7/N] Make FSDP support NamedOptimizer (#91160)

fegin committed 1 year ago

6cea4f3d

[vision hash update] update the pinned vision hash (#91284)

pytorchmergebot committed 1 year ago

71318742

Restore RNG states for composable reentrant activation checkpointing (#91265)

mrshenli committed 1 year ago

a0554261

Run test_spectral_ops serially to fix CUDA illegal memory access (#91264)

huydhn committed 1 year ago

8f165245

Fix non-existing parameters in docstrings in torch/distributed (#91116)

kit1980 committed 1 year ago

365071c7

Remove inductor performance from ciflow/nightly as infra is not ready to handle these jobs… (#91271)

weiwangmeta committed 1 year ago

b50f379c

use scatter_add for index_add when dim is the most inner dim (#88729)

CaoE committed 1 year ago

68e9da68

add mixed data type support for GroupNorm backward on CPU (#88663)

CaoE committed 1 year ago

59a5be3b

add cu118 workflows for Windows (#91216)

ptrblck committed 1 year ago

8e55d583

[MPS] Fix the error with high watermark value on x86 (#91268)

razarmehr committed 1 year ago

014d7802

Fix for RNN/LSTM/GRU modules to work with stateless.functional_call() (#91111)

jbschlosser committed 1 year ago

85e393ba

[Vulkan] Use EXPECT_EQ instead of ASSERT_TRUE in vulkan_api_test querypool_flushed_shader_log (#91259)

salilsdesai committed 1 year ago

300f7777

[quant] Support lowering for quantized embedding byte operator (#91159)

jerryzh168 committed 1 year ago

2a23dfe8

Revert "Store source, not sname, in Symbol (#91057)"

pytorchmergebot committed 1 year ago

b68fd7e3

[Resubmit] Require inductor to match stride order (#91185)

eellison committed 1 year ago

6e0cd8b9

[FSDP][optim_state_dict][6/N] Refactor the optim_state_dict APIs to support hooks (#90798)

fegin committed 1 year ago

1ab6ac46

Newer Older