strisunshinewentingwang/pytorch

Pull Requests Commits

Retry - [NVFuser] always use fallback if fusion fails

davidberard98 committed 2 years ago

e52dc988

Add torch.linalg.ldl_factor_ex and torch.linalg.ldl_solve

IvanYashchuk committed 2 years ago

8bb72030

[ONNX] Documentation and typing annotations in registry

justinchuby committed 2 years ago

2e841e68

Readme update to remove old python version

albanD committed 2 years ago

8880e3a3

Add test_jit_cuda_fuser.py to NVFuser merge rules

davidberard98 committed 2 years ago

5b6efb82

Nvfuser - Type Promotion Fix

rdspring1 committed 2 years ago

e9f17da2

[NVFuser] Reduce accuracy threshold for fp16 nvfuser opinfos

davidberard98 committed 2 years ago

4d527cd2

Shard rocm distributed tests

janeyx99 committed 2 years ago

ad4f51d7

Disable XLA test workflow

janeyx99 committed 2 years ago

833d65ae

Add __all__ for torch.cuda.memory

anjali411 committed 2 years ago

9bf2c87e

Hipify fixes for a successful DeepSpeed build

rraminen committed 2 years ago

7422ccea

[quant][core][gpu][feature] Implemented quantized cuda adaptive average pool2d op (#76081)

dzdang committed 2 years ago

bbc263eb

Adding `broadcast_in_dim` and non-contiguous Tensor support NVFuser Python Frontend

kevinstephano committed 2 years ago

f51516df

[PT-D][Sharding] Move Partial Tensor to the _shard folder and add logic to remove padding (#76199)

fduwjj committed 2 years ago

c1037d0d

[quant] Fix tensorrt config after the backend_config_dict refactor (#76414)

jerryzh168 committed 2 years ago

b69d44da

[shard] Extensible Sharder and ShardingPlanner (#75844)

wanchaol committed 2 years ago

a37141ad

Revert "[JIT] Register decomp reland"

pytorchmergebot committed 2 years ago

a5bc02ae

Revert "[JIT] Add another decomposition api"

pytorchmergebot committed 2 years ago

2af0c42a

[quant][core][gpu][improvement] Converted reinterpret_cast<T *>(some_int8_tensor.data_ptr()) calls to some_int8_tensor.data_ptr<int8_t> in quantized cudnn operator files (#75980)

dzdang committed 2 years ago

9e73e890

Make lazy tensor ptr class customizable (#76476)

wconstab committed 2 years ago

d0cb31d5

Make lazy tensor creation and value strings customizable (#76472)

wconstab committed 2 years ago

4cae5708

Fix GenLazyIR.node_base_ctor_call (#76471)

wconstab committed 2 years ago

cfc90cf3

[RPC small change] Improve logging from 'unknown destination worker'

H-Huang committed 2 years ago

ec420764

[quant][core][gpu][improvement] Made exception throwing message clearer in Quantized Cudnn Conv2d, Linear, and Add ops (#76102)

dzdang committed 2 years ago

15884418

[quant][core][gpu][feature] Added support for float->quantized cuda tensor copying (#76177)

dzdang committed 2 years ago

ad88816c

[JIT] Add another decomposition api

Elias Ellison committed 2 years ago

32cb0375

Decouple LTC from TS Backend using Lazy IR Builder

antoniojkim committed 2 years ago

f3f327e1

Revert "Revert "Allow specifying tags for aten operators in native_functions.yaml""

anjali411 committed 2 years ago

b204ad86

[FX] Fix operator_schemas normalize_function to consider OpOverloads (#76469)

jfix71 committed 2 years ago

1c5a66c2

[quant][gpu][core][bug fix] Added memset to CacheKey for quantized cudnn linear op (#76447)

dzdang committed 2 years ago

b7bd677e

Newer Older