strisunshinewentingwang/pytorch

Pull Requests Commits

Enable torch._C._get_privateuse1_backend_name in Dynamo tracing (#103141)

heidongxianhua committed 1 year ago

c24b61bc

[inductor] add the ability to do heavier search for coordinate descent tuning (#99403)

shunting314 committed 1 year ago

6095a22c

Deprecate "Type" and support more devices for save_on_cpu (#103245)

heidongxianhua committed 1 year ago

2961ea80

Debug Windows locked files (#103237)

huydhn committed 1 year ago

c037088a

[dtensor] support torch.save/load with DTensor (#103106)

wanchaol committed 1 year ago

4cc474de

Get rid of dim_groups attribute from DeviceMesh (#103105)

wanchaol committed 1 year ago

d31707a2

numpy1.25 deprecation: np.product -> np.prod (#103263)

ppwwyyxx committed 1 year ago

81b704ea

ImportLib py3.10 bug in AOTInductor (#103277)

msaroufim committed 1 year ago

f3553c50

[dashboard] Allocate 4 shards for torchbench (#103280)

desertfire committed 1 year ago

4c03adc1

add github action to upload alerts to rockset / aws (#102995)

PaliC committed 1 year ago

8c584028

[FSDP][Easy] Remove redundant var def in test (#103270)

awgu committed 1 year ago

bb827873

inductor: fix bf16 legalization issue for fp32 load with to bf16 case (#103080)

XiaobingSuper committed 1 year ago

8e5b7ce5

Update error message with torch logging instructions (#102892)

mlazos committed 1 year ago

40dbbcab

[Specialized Kernel] Translate Kernel Assignment Logic from function.yaml to native_functions.yaml (#102576)

Jack-Khuu committed 1 year ago

d0c0e13b

Put back cuda 11.8 distributed tests (#103265)

atalman committed 1 year ago

98a1e3a3

add huggingface to inductor docker images (#102881)

PaliC committed 1 year ago

481023fb

[quant][pt2] Fix convert in Conv + BN + ReLU QAT fusion (#102993)

andrewor14 committed 1 year ago

89d57f26

Dropout support for memory efficient attention (#102038)

drisspg committed 1 year ago

606fb882

Manually generate guards for optimizer (#103121)

mlazos committed 1 year ago

05e91a50

[FSDP] Reshard frozen params in backward (#101982)

awgu committed 1 year ago

48056b16

Revert "Add print statements to debug sharding error (#102713)"

pytorchmergebot committed 1 year ago

b52ee80c

Add early validation logic to dynamic_dim (#102982)

tugsbayasgalan committed 1 year ago

cea899cd

Fix GELU-related docstring formatting (#102845)

MysticalMusings committed 1 year ago

f1f13a35

[ROCM] enable hipSOLVER backend for linalg.ldl_factor (#102665)

xinyazhang committed 1 year ago

1d857586

[Dynamo Hackathon] Add support for hasattr on TorchVariable (#103177)

gmagogsfm committed 1 year ago

b4f3a6f5

[cuda] Limit grid size for torch.cat kernel on aligned16 contig tensors (#103233)

valentinandrei committed 1 year ago

c62fcedc

Make dynamo bench conditionally import DDP/FSDP (#103163)

wconstab committed 1 year ago

39201ce0

[CI] Enable UCC in CI (#100395)

Aidyn-A committed 1 year ago

591134f2

Rename READEME.md to README.md (#103230)

kit1980 committed 1 year ago

a1c26ba7

[dynamo] Fix Autograd Function Classmethod bug (#103175)

angelayi committed 1 year ago

4a72708d

Newer Older