strisunshinewentingwang/pytorch

Pull Requests Commits

Dont clone inputs if using fake tensor (#88208)

eellison committed 1 year ago

60925fcb

[Pytorch][vulkan] Generate shader with parameters (#88322)

kimishpatel committed 1 year ago

192e806c

[minor] use set_default_dtype instead of try and finally (#88295)

kshitij12345 committed 1 year ago

fe3a226d

[dashboard] Replace aot_nvfuser with nvprims_nvfuser (#88437)

anijain2305 committed 1 year ago

f8b73340

[Dynamo][Easy] Fix config.suppress_errors error log (#88402)

yanboliang committed 1 year ago

2bda2baa

Verbose exc printing fix (#88387)

mlazos committed 1 year ago

4d62ee1b

[ONNX] Default runtime type checking to raising errors (#86555)

justinchuby committed 1 year ago

0a274c4b

add parameters check for mkldnn_transpose (#85318)

XiaobingSuper committed 1 year ago

d70bc222

[dynamo] Support compare op for userfunctionvariable (#88372)

anijain2305 committed 1 year ago

c1dd13fb

Disallow module attribute mutation (#88354)

mergennachin committed 1 year ago

2c46d572

Revert "Fix primTorch compute_elementwise_output_strides (#88175)"

pytorchmergebot committed 1 year ago

2b117c84

disable the out variants in test_cumprod test for inductor (#88328)

Krovatkin committed 1 year ago

0f6304ef

add an exclude for test_constructor for inductor (#88143)

Krovatkin committed 1 year ago

529ba076

better error message for out= ops (#88367)

Krovatkin committed 1 year ago

002dad35

reduce the number of autotuning iterations, don't autotune simple til… (#88386)

ngimel committed 1 year ago

b4fcfe77

Add support for neg to NestedTensor (#88131)

cpuhrsch committed 1 year ago

5e6ceebc

[FSDP()][Easy] Make `fully_shard()` only `FULL_SHARD` (#88260)

awgu committed 1 year ago

35be73df

[FSDP()] Have `fully_shard()` abide by `@contract`! (#88235)

awgu committed 1 year ago

fc743ec0

Add a shortcut in Makefile for updating triton (#88318)

desertfire committed 1 year ago

63cd5d7e

Make Python op registration work with torchdeploy/multipy (#87162)

ezyang committed 1 year ago

f884e817

Add a reshape_copy operator. (#88314)

ezyang committed 1 year ago

2f296cfd

Put Python Dispatcher cache in dict, clear it on new registrations. (#88329)

ezyang committed 1 year ago

86c7cd28

Unconditionally enable python dispatcher in AOTAutograd (#88365)

ezyang committed 1 year ago

97d3b200

[FSDP] Do not include empty state in `_flatten_optim_state_dict()` (#88353)

awgu committed 1 year ago

a6895022

[FSDP()][Easy] Rename `_State` to `_FSDPState` (#88234)

awgu committed 1 year ago

95a9721a

[FSDP()] Rename to `fully_shard()` and move to `_composable/` (#88233)

awgu committed 1 year ago

0520131e

[fix] allow saving python attr on Tensor and Parameter via torch.save (#81616)

kshitij12345 committed 1 year ago

54b6188c

Fix primTorch compute_elementwise_output_strides (#88175)

SherlockNoMad committed 1 year ago

1c8a0656

Make GenLazyNativeFuncDefinition generator to be customizable in lazy codegen (#87823)

wonjoolee95 committed 1 year ago

0efd4e92

Update all ONNX symbolics with new JitScalarType API (#87245)

Thiago Crepaldi committed 1 year ago

a8f40b39

Newer Older