strisunshinewentingwang/pytorch

Pull Requests Commits

merge identical branches in cpu_index_kernel (#102601)

cyyever committed 1 year ago

850b37cc

Fix ignored_states when they are passed as generators (#102575)

zhaojuanmao committed 1 year ago

f47ee877

Add integer overflow checks for large compressed tensor dimensions and nnz (#102530)

pearu committed 1 year ago

9f97b7c4

Improve repeat_interleave with scalar repeat value (#102570)

peterbell10 committed 1 year ago

9fd14fcd

[inductor] Revert a CI remedy for Triton compilation error (#102541)

desertfire committed 1 year ago

47b884a7

nn.Linear with BSR inputs: spare the user from explicit Triton kernel registrations (#98403)

nikitaved committed 1 year ago

d80d3b18

[Executorch][codegen] Add ETKernelIndex for aggregating all kernels for kernel (#102565)

larryliu0820 committed 1 year ago

019c3862

apply constexpr and if constexpr when possible (#102471)

cyyever committed 1 year ago

7c2641d5

Quick fix for keep-going + reruns (#102569)

clee2000 committed 1 year ago

a5ddb72a

[inductor] Fix issue with 0D reductions (#102568)

jansel committed 1 year ago

3c0251a1

[inductor][pattern matcher] Retain meta tags (#102462)

anijain2305 committed 1 year ago

46691d43

Add dynamo collections.deque support (#102412)

wconstab committed 1 year ago

e7cc4177

Fix warning couldn't find split args (#102561)

devashishshankar committed 1 year ago

cdca25cd

[ONNX] Reduce exporter memory usage by removing intermediate values (#101148)

ilyasher committed 1 year ago

b4a49124

[profiler] Reintroduce forward-backward links (#102424)

davidberard98 committed 1 year ago

5324124e

add function specializations for the case of parameters in BFloat16 data type (#100233)

CaoE committed 1 year ago

73fd7235

[build] fix compilation error on s390x (#101923)

kiszk committed 1 year ago

9edf65a8

[MPS] Fix softplus with f16 input (#101948)

qqaatw committed 1 year ago

cce58a43

[dynamo][higher order op] Bugfixes to pass graph.lint (#102448)

anijain2305 committed 1 year ago

c3c14961

[ONNX] Don't duplicate model weights in ONNX export (#101134)

ilyasher committed 1 year ago

8d5b14e9

[benchmark] Flag to switch on activation checkpointing for HF models (#102557)

anijain2305 committed 1 year ago

33a49eea

Switch cuda 12.1 docker images to gcc9 (#102380)

atalman committed 1 year ago

6ac8a117

[Dynamo] Save global autocast state to restore on graph break (#102415)

yanboliang committed 1 year ago

9ff1932d

Revert "Quick fix for keep-going + reruns (#102569)"

pytorchmergebot committed 1 year ago

1a6ab8a5

[PT2][Quant][BE] refactor tets cose to reduce duplication and standardize (#102497)

kimishpatel committed 1 year ago

4f468646

Quick fix for keep-going + reruns (#102569)

clee2000 committed 1 year ago

7f6edcf4

[quant][pt2e] Add support for FixedQParamsQuantizationSpec (#102439)

jerryzh168 committed 1 year ago

f14ac74f

[fx] Fix repr when arg is an OpOverload (#102547)

anijain2305 committed 1 year ago

168ae806

[minifier] add missing import (#102521)

anijain2305 committed 1 year ago

68e55bff

Revert "[pt2] add `SymInt` support for `linalg.tensorsolve` (#102466)"

pytorchmergebot committed 1 year ago

95cdd58c

Newer Older