strisunshinewentingwang/pytorch

Pull Requests Commits

pt2e short term quant: respect qmin/qmax for linear weight

vkuzo committed 1 year ago

e821eb86

[7/N][BE] Remove Partial Tensor and its dependency (#95949)

fduwjj committed 1 year ago

28aa2efd

[6/N][BE] Remove Sharded Linear Op for ShardedTensor (#95948)

fduwjj committed 1 year ago

6dddc0d6

[5/N][BE] Remove Replicated Tensor class (#95947)

fduwjj committed 1 year ago

4e396a54

Revert "memory viz: Add colors for categories and a legend (#90587)"

seemethere committed 1 year ago

b38b39c4

Use original arg names if possible (#95898)

voznesenskym committed 1 year ago

22c9896e

Delete torch._functorch.config.use_dynamic_shapes (#96102)

ezyang committed 1 year ago

6fff2322

[CI] Use CUDA 11.8 to run inductor benchmark tests (#96059)

desertfire committed 1 year ago

483e193d

Use working pre-built sccache binary (#95997)

huydhn committed 1 year ago

000d9ec1

Only print bytecode of inlined function if output_code is True (#95969)

ezyang committed 1 year ago

69bfdcd2

fix typo in comments under torch/csrc/autograd (#96061)

kiszk committed 1 year ago

69aa6b4b

[primTorch] move diagonal & add linalg.diagonal refs (#95774)

khushi-411 committed 1 year ago

301a28bf

Update skips for RecursionError (#96109)

ezyang committed 1 year ago

1fd7ea1b

Multiple fixes for functional collectives. (#95897)

Rodrigo Kumpera committed 1 year ago

5b2ab0dd

USE_FAST_NVCC Windows (#95206)

mantaionut committed 1 year ago

3beafc91

dynamo: wrap graph break inst in try except block - with context manager setup/teardown (#94758)

jon-chuang committed 1 year ago

7a192cc5

sparse compressed tensor validation without syncs for low-(batch)dim tensors. (#94048)

nikitaved committed 1 year ago

18d6ce96

use float to init reduction value (#95452)

zhuhaozhe committed 1 year ago

ebaf9af7

inductor: pre-convert a TensorBox's layout to FixedLayout at FX side if one user of it is a CPU external customer kernel (#95873)

XiaobingSuper committed 1 year ago

dcc159d3

Upload torch dynamo perf stats to Rockset (#95675)

huydhn committed 1 year ago

cc775fb8

[CI] Make inductor-perf-test-nightly produce data for dashboard (#95685)

desertfire committed 1 year ago

02792ff1

Error when jit.trace/script is used with torch.compile (#91681)

msaroufim committed 1 year ago

fa92b6a7

Fix node provenance tracking (#95901)

Chillee committed 1 year ago

e8cd173a

Automated submodule update: kineto (#95798)

facebook-github-bot committed 1 year ago

36a6e2c5

[inductor] Add some simple decomps (#96039)

jansel committed 1 year ago

5dd52e25

[CI] Skip xcit_large_24_p8_224 in TIMM (#96048)

desertfire committed 1 year ago

60cf9561

[CI] Further tighten the checking of two eager runs (#95902)

desertfire committed 1 year ago

1359d16f

[inductor] use triu ref instead of lowering (#96040)

ngimel committed 1 year ago

43e71cdd

[dtensor] refactor shape/offset calculation (#95923)

wanchaol committed 1 year ago

789fc4c2

Fix training enablement in AOTAutograd (#95975)

ezyang committed 1 year ago

af8dbe7e

Older