strisunshinewentingwang/pytorch

Pull Requests Commits

Symintify numel(), infer_size, prims.elementwise_meta (#88956)

SherlockNoMad committed 2 years ago

caf3d531

Add support for dynamic kwarg to torch._dynamo.optimize (#89290)

ezyang committed 2 years ago

7c811efa

Revert "Symintify numel(), infer_size, prims.elementwise_meta (#88956)"

pytorchmergebot committed 2 years ago

8ad39536

Add nullptr_t overload to c10::intrusive_ptr (#89196)

kvathupo committed 2 years ago

8ac58bc2

Reland 2 "Towards unifying symbolic and non symbolic fake tensor (#89038) (#89143)" (#89346)

ezyang committed 2 years ago

5582001b

[PT-D][1/N] Sync TP Beta change to prod (#89242)

fduwjj committed 2 years ago

6afe3412

RM expectedFailure UnspecReproTests.test_batch_norm_act_unspec (#89340)

voznesenskym committed 2 years ago

6b8c1b19

[ONNX] Add setType from user into InferredType and Reliable in ConstantValueMap (#88622)

titaiwangms committed 2 years ago

6daf60be

[quant][fix] Add quant_min/quant_max for default dynamic quantization observer (#89267)

jerryzh168 committed 2 years ago

940959eb

Fix try/except flow where DataDependentOutputException is getting wrapped in a RuntimeError (#89314)

voznesenskym committed 2 years ago

808bdbab

Added utility to count memory reads/written in Inductor (#89203)

Chillee committed 2 years ago

419ef2cd

add jvp test with non-contig inputs (#89131)

kshitij12345 committed 2 years ago

7a2930b3

Add --explain flag to bench (#89316)

voznesenskym committed 2 years ago

631baecb

Don't redefine __STDC_FORMAT_MACROS (#89310)

ppwwyyxx committed 2 years ago

e6996ea1

cast C++ py-bound SymNode to SymInt correctly (#89295)

Krovatkin committed 2 years ago

8c0515db

Update sdp dispatch logic to enable fused backward (#89154)

drisspg committed 2 years ago

2e72ec79

[dynamo] mutable local caching to make dynamo faster at tracing mutation (#89170)

mlazos committed 2 years ago

85a87e63

Move bazel to c++17 (#89297)

malfet committed 2 years ago

ea58955d

[dashboard][huggingface] skip accuracy checks for really large models… (#89273)

anijain2305 committed 2 years ago

cad5772c

[small] Update error message (#89294)

H-Huang committed 2 years ago

ee907375

[functorch] introduce an experimental map() op. (#88767)

zhxchen17 committed 2 years ago

c3938bb9

Detach fake tensors into val, so they aren't affected by metadata mutation (#89140)

ezyang committed 2 years ago

94b5c807

[BE] Print backtraces from coredumps (#89309)

malfet committed 2 years ago

885f8a56

[FX] Add type annotation to `getitem` node before `split_module` (#88510)

lequytra committed 2 years ago

0e1fcc8a

[Inductor CI] Use string format for cuda-arch-list input to prevent 8.0/9.0/10.0 etc from being interpreted as 8/9/10 (#89279)

weiwangmeta committed 2 years ago

ecfb4e06

Add NVTX markers that dump additional information for nvprim_nvfuser Dynamo graphs (#88259)

blongnv committed 2 years ago

7551136b

[Profiler] Don't raise SOFT_ASSERT in debug builds. (#89240)

Taylor Robie committed 2 years ago

35d5fc52

[DDP][Docs] Add warning that `no_sync()` should include forward (#89244)

awgu committed 2 years ago

bfffc8d8

Re-enable test_hf_bert_fsdp (#89223)

davidberard98 committed 2 years ago

304b5de1

Don't trace when we track_tensor_tree (#89139)

ezyang committed 2 years ago

ba605c3b

Newer Older