strisunshinewentingwang/pytorch

Pull Requests Commits

Fix DLPack CUDA stream convention (#67618)

Emilio Castillo committed 2 years ago

533e72e0

[testing] make @dtypes mandatory when using @dtypesIf (#68186)

kshitij12345 committed 2 years ago

d5d2096d

torch.linalg.qr: forward AD support (#67268)

nikitaved committed 2 years ago

857fed1f

[BE] MapAllocator: report map error on Linux (#68545)

malfet committed 2 years ago

a2d187a6

Fix `_make_wrapper_subclass`'s storage_offset handling (#68268)

zou3519 committed 2 years ago

b1aa45a8

Stop warning spamming about vmap in gradcheck (#68586)

zou3519 committed 2 years ago

f0e2ad50

Replace empty with new_empty in nn.functional.pad (#68565)

zou3519 committed 2 years ago

f9ef807f

[quant][embedding qat] eager mode QAT for Embeddings (#66429)

b-koopman committed 2 years ago

6c9cf5e6

[GPU host alloc] Fast path for size 0 malloc (#68532)

jaewonlee-fb committed 2 years ago

dbbb0247

[static runtime][dper] multi_env tests for static runtime: selective enable (#67467)

ajyu committed 2 years ago

4635f571

[reland] simplify init_from_local_shards API (#68021)

wanchaol committed 2 years ago

35712a8e

Sparse CSR: add `convert_indices_from_csr_to_coo` (#66774)

rok committed 2 years ago

952ca25d

Fix c10d TCP store with mutex (#68499)

jiayisuse committed 2 years ago

96ba2099

Enable desync root cause analysis for NCCL (#68310)

jiayisuse committed 2 years ago

146a7f68

`scatter_reduce` (#68115)

rusty1s committed 2 years ago

98077871

[fx2trt] add converter for acc_ops.hardtanh (#68550)

842974287 committed 2 years ago

e72b9db4

[predictor][trt] Expose more CUDA/CuDNN info to at::Context and BC stage 1 (#68146)

kflu committed 2 years ago

9d9ca88f

[android][fbjni] Update fbjni to 0.2.2 (#68400)

IvanKobzarev committed 2 years ago

d71092f6

[bugfix] TensorList args in functionalization pass (#68395)

bdhirsh committed 2 years ago

53bfb00e

[ONNX] Release values cached in global object (#68210)

wschin committed 2 years ago

b0bdf588

Refactor saving jit::Module to mobile .pt in 2 steps: (#66494)

qihqi committed 2 years ago

4eb772fd

Improve native layer norm backward perf (#68238)

ngimel committed 2 years ago

e2aeb4a7

Actually enable PYTORCH_RETRY_TEST_CASES for linux tests (#68486)

janeyx99 committed 2 years ago

f3e2fefe

[quant][graphmode][fx] Refactor node_name_to_target_dtype to make it more clear (#68317)

jerryzh168 committed 2 years ago

2f37a39a

Remove TH/THC Storage data and copy functions (#68127)

kurtamohler committed 2 years ago

3b4f0723

Use TORCH_CHECK in MapAllocator (#68424)

peterbell10 committed 2 years ago

4e21d77d

docs: Added Union to supported types in documentation (#68435)

frgfm committed 2 years ago

693fe2fd

[SR] Add StorageGroup abstraction (#68279)

Mike Iovine committed 2 years ago

61206ba4

add torch.diff support for n greater than 1 (#67260)

mikaylagawarecki committed 2 years ago

cac3cd14

Added antialias flag to interpolate (CPU only, bilinear) (#65142)

vfdev-5 committed 2 years ago

3da2e09c

Newer Older