strisunshinewentingwang/pytorch

Pull Requests Commits

[DataPipe] Adding BatcherMapDataPipe (#68197)

NivekT committed 2 years ago

0465f64b

Revert D32010095: [pytorch][PR] Add ability for a mobile::Module to save as flatbuffer

albanD committed 2 years ago

00ebbd5e

[Static Runtime] Skip ProcessedNode:: verify_no_memory_overlap() for out variants (#68639)

Hao Lu committed 2 years ago

ed3b73fd

[shard] add back init_from_local_shard_and_global_metadata API (#69226)

wanchaol committed 2 years ago

c60232d8

support pure fp16 training in FSDP (#68417)

zhaojuanmao committed 2 years ago

12621c3a

Add ability for a mobile::Module to save as flatbuffer (#67351)

qihqi committed 2 years ago

41d35dc2

[JIT] Compute input sym shapes in partial eval graph (#68281)

Elias Ellison committed 2 years ago

40fb28ea

[DataPipe] Simplify BatcherIterDataPipe by removing 'unbatch_level' argument and functionality (#68594)

NivekT committed 2 years ago

d8a44270

[deploy] docs (#69251)

suo committed 2 years ago

ad182479

Back out "[CUDA Pinned Memory] Event recording with non-blocking copies should track the storage context, not the tensor data pointer" (#69193)

dstaay-fb committed 2 years ago

cbe0a38d

Back out "[CUDA Pinned Memory] Alternative implementation of pinned memory allocator focusing on multi-threaded scalability" (#69191)

dstaay-fb committed 2 years ago

929f2a75

Strided masked var. (#68738)

pearu committed 2 years ago

370d0afc

[Pytorch Edge] Update Black Box Api with operator versioning (#68678)

JacobSzwejbka committed 2 years ago

291e56ed

[Operator Versioning][Edge] Add old models and unittest (#67726)

cccclai committed 2 years ago

b9738e92

RegisterDispatchKey.cpp: remove redundant code (#68983)

guoyejun committed 2 years ago

124bb6a1

[torch][distributed] Check for file existence before invoking cleanup logic in FileStore destructor (#68603)

aivanou committed 2 years ago

fced51ea

fixing layer_norm cuda bug (#69210)

jjsjann123 committed 2 years ago

3c1e2ff9

[pyper] add flag to disable clip_ranges_gather fusions (#69198)

ajyu committed 2 years ago

d72d4768

Fix RAdam docstring on LR default value (#69186)

bryant1410 committed 2 years ago

263125a9

Change misleading MaxUnpool2d example to better demonstrate output_size usage (#68936)

kylematoba committed 2 years ago

3bf4080f

add `extra_repr` for `nn.ZeroPad2d` (#69206)

peterjc123 committed 2 years ago

2eef5e76

Revert D32329330: [JIT] Separate GPU implementation of frozen_conv_add_relu_fusion.cpp

davidberard98 committed 2 years ago

cd043c33

[LTC] Upstream helpers for c10::Device <=> BackendDevice (#69064)

Jiewen Tan committed 2 years ago

e6c435bf

remove accidentally committed redundant debug print (#68510)

zhaojuanmao committed 2 years ago

92f16894

Strided masked normalize. (#68694)

pearu committed 2 years ago

1842364b

record the datapipe for each pieces of Dataset (#67613)

guotuofeng committed 2 years ago

23633bdb

Add kl divergence between normal and laplace distribution. (#68807)

paulstreli committed 2 years ago

deaf745a

Dataset & IterableDataset attribute errors prints attribute (#69021)

TSPoulsen committed 2 years ago

486ae5c7

Check that block height and width are positive in `nn.Fold` (#69048)

kurtamohler committed 2 years ago

d507fd63

Introduce `IS_LINUX` and `IS_MACOS` global vars (#69093)

malfet committed 2 years ago

c08e95dd

Newer Older