strisunshinewentingwang/pytorch

Pull Requests Commits

Add support for MatMul to BatchMatMulFP16Acc{16,32}Fake Op Mapping

venkatacrc committed 3 years ago

580831bf

[ONNX] Update ONNX to rel-1.9 (#55889) (#57080)

neginraoof committed 3 years ago

599f5058

.github: Remove windows dependency installs (#59283)

seemethere committed 3 years ago

f87aa231

[reland] Make TP agent use streams from Future when sending response (#59212)

lw committed 3 years ago

3a2149a4

[reland] Set and propagate devices in RRef completion future (#59211)

lw committed 3 years ago

258a9910

[reland] Set streams when invoking UDFs (#59210)

lw committed 3 years ago

a3392caf

[reland] Create CUDA-aware futures in RequestCallback (#59209)

lw committed 3 years ago

f8a3fd4e

[reland] Provide pre-extracted DataPtrs when completing a Future with a Message (#59208)

lw committed 3 years ago

3af6ff98

[reland] Allow Future::then to return pre-extracted DataPtrs (#59207)

lw committed 3 years ago

1adc289e

[reland] Always use intrusive_ptr for Message (2 out of 2) (#59206)

lw committed 3 years ago

b07d68e2

[reland] Always use intrusive_ptr for Message (1 out of 2) (#59205)

lw committed 3 years ago

5ec169b4

Alias for `i0` to `special` namespace (#59141)

krshrimali committed 3 years ago

44c20ce6

Fix breakpad build and add to more images (#59236)

driazati committed 3 years ago

059a717c

[RPC Framework] Support creating a RemoteModule by RRef (#59242)

Yi Wang committed 3 years ago

dbe629c5

[quant][graphmode][fx][fix] Fix support for custom module (#59041)

jerryzh168 committed 3 years ago

3218d890

[quant][graphmode][fx][refactor] Remove Quantizer class from convert (QuantizeHandler) (#59040)

jerryzh168 committed 3 years ago

06af7618

fix numpy compatibility in test for `torch.kthvalue` (#59214)

pmeier committed 3 years ago

0a267819

Fix device of info tensor for torch.linalg.inv_ex with MAGMA backend (#59223)

IvanYashchuk committed 3 years ago

e9e1bb1a

[quant][graphmode][fx][refactor] Remove Quantizer class from quantize_node (#59039)

jerryzh168 committed 3 years ago

50e6ee3c

CUDA support in the CSR layout: CUDA addmm/matvec (#59012)

aocsa committed 3 years ago

2d8f0d96

[CUDA graphs] Makes sure all graphs tests call empty_cache() at some point before capture (#59233)

mcarilli committed 3 years ago

3efefc40

[quant][graphmode][fx][refactor] Remove _prepare from Quantizer class (#59038)

jerryzh168 committed 3 years ago

1d37f415

[Reland] Adds an aten::_ops namespace with unambiguous function names (#59018)

zou3519 committed 3 years ago

970096b6

use long index type for index_add_cuda deterministic path (#59254)

yuguo68 committed 3 years ago

8805093e

[quant][graphmode][fx][refactor] Remove find_matches from Quantizer class (#59037)

jerryzh168 committed 3 years ago

20348fb3

[quant][graphmode][fx][refactor] Remove fold_weights from Quantizer class (#59036)

jerryzh168 committed 3 years ago

7d64fc67

DOC Adds register_module_full_backward_hook into docs (#58954)

thomasjpfan committed 3 years ago

8af62812

[nnc] Enable CPU fusion inside Facebook, take 3 (#59253)

bertmaher committed 3 years ago

6e7dae9c

[quant][graphmode][fx][refactor] Remove save_state and restore_state from Quantizer class (#59035)

jerryzh168 committed 3 years ago

cc489180

Fix mismatch in README.md Docker Image section (#59199)

ChenYuHo committed 3 years ago

336ac949

Newer Older