Fix error message when using as_strided with negative striding good first issue topic: error checking triaged

#33290 opened Feb 13, 2020 by daniilhayrapetyan

2

Python/C++ API Parity: torch.nn modules and functional good first issue module: cpp module: nn triaged

#25883 opened Sep 9, 2019 by yf225

99

Fuse operators inside of the (CPU) glu operator. good first issue module: cpu topic: bootcamp triaged

#26418 opened Sep 18, 2019 by VitalyFedyunin

4

LogCumsumExp feature good first issue module: operators topic: bootcamp topic: operator triaged

#26411 opened Sep 18, 2019 by agadetsky

16

Remove unpack() in torch/csrc/nn/type_checks.h and its caller functions in the codebase good first issue module: internals module: nn triaged

#20704 opened May 20, 2019 by yf225

4

[onnx] Silent crash when exporting a model larger than 2gb good first issue module: onnx triaged

#19277 opened Apr 15, 2019 by gkossakowski

13

Cumulative Maximum feature good first issue module: operators topic: bootcamp topic: operator triaged

#20240 opened May 7, 2019 by cannon

36

new_zeros is not traced correctly good first issue jit

#16663 opened Feb 1, 2019 by houseroad

6

Inconsistent behaviour of torch.fmod on CPU and GPU good first issue module: cuda triaged

#15921 opened Jan 10, 2019 by nikitautiu

16

Add an argument 'beta' in smooth L1 loss? good first issue

#16357 opened Jan 25, 2019 by wangg12

8

[docs] Strange order of items in docs contents in left pane module: docs triaged

#33055 opened Feb 6, 2020 by vadimkantorov

3

nn.LSTM output shape is mis-documented module: docs triaged

#33215 opened Feb 12, 2020 by LiboShen

2

PyTorch C++ API docs only tracks master branch module: cpp module: doc infra triaged

#32237 opened Jan 15, 2020 by yf225

Documentation is not loaded by IDEs module: docs triaged

#32767 opened Jan 29, 2020 by stefanopini

Document different gradient wrt to jax when nesting module: autograd module: docs triaged

#32576 opened Jan 24, 2020 by MikeInnes

13

Missing documentation for is_quantized quantization triaged

#32064 opened Jan 10, 2020 by raghuramank100

1

[docs] Missing docs for torch.no_grad, instead there are docs for torch.autograd.no_grad actionable fixathon module: autograd module: docs triaged

#32014 opened Jan 9, 2020 by vadimkantorov

1

Exclude master documentation from google module: doc infra triaged

#31894 opened Jan 6, 2020 by zou3519

Missing edge case information in BCELoss documentation high priority module: docs topic: loss triaged

#31453 opened Dec 19, 2019 by Enealor

6

[docs] torch.histc docs mention min/max have int type module: docs triaged

#31475 opened Dec 19, 2019 by vadimkantorov

4

kl_div formula in documentation module: docs module: nn triaged

#32278 opened Jan 16, 2020 by ThyrixYang

2

Max-pooling uses implicit negative infinity padding, not zero-padding as indicated in documentation triage review triaged

#33384 opened Feb 15, 2020 by josh-gleason

1

Add Tensor.nbytes() / `Tensor.itemsize()` good first issue low priority module: numpy triaged

#12728 opened Oct 16, 2018 by ezyang

5

Docs missing link to torch.__config__ module: docs triaged

#30384 opened Nov 24, 2019 by danieljbrooks

Typo in `torch.utils.tensorboard.add_image` visualization

#31603 opened Dec 25, 2019 by kdbanman

Documentation for `scatter` incorrectly states that index values must be unique module: docs module: operators topic: bc-breaking triaged

#31776 opened Jan 2, 2020 by ngimel

6

[docs] F.ctc_loss docs to warn clearly about invalid inf-causing inputs; zero_infinity to become enabled by default module: docs triaged

#31557 opened Dec 22, 2019 by vadimkantorov

13

Need to add documentation for c10d.ProcessGroupGloo and c10d.ProcessGroupNCCL better-engineering module: distributed module: docs triaged

#31036 opened Dec 10, 2019 by pritamdamania87

1

[docs] Improve docs of nn.MultiheadAttention module: docs triaged

#28721 opened Oct 25, 2019 by vadimkantorov

3

Document memory characteristics of in-place ops enhancement module: docs module: sparse topic: type promotion triaged

#30533 opened Nov 27, 2019 by nairbv

1

Fix softplus clampling issues using logsigmoid module: nn triaged

#31856 opened Jan 4, 2020 by joaompereira

Unexpected Behavior when Pointwise Operations Write to Expanded Tensors good first issue module: internals topic: partial aliasing triaged

#10756 opened Aug 21, 2018 by c0nn3r

5

random_pruning, remove_pruning typo module: docs triaged

#32991 opened Feb 4, 2020 by vainaijr

2

Inconsistent documentation for in-place functions at https://pytorch.org/docs/stable/torch.html module: docs triaged

#29028 opened Nov 1, 2019 by shubho

7

torch.from_file not documented module: docs triaged

#28794 opened Oct 28, 2019 by colesbury

1

[docs] Unclear input/output format for TransformerEncoderLayer module: nn triaged

#28657 opened Oct 25, 2019 by vadimkantorov

9

Stable docs show dead __config__ section link module: docs triaged

#28110 opened Oct 16, 2019 by daniel-j-h

2

Warnings in the docs build should be errors in CI module: doc infra triaged

#27765 opened Oct 11, 2019 by zou3519

1

Ability to download docs HTML for offline use module: docs triaged

#31321 opened Dec 16, 2019 by samlobel

2

Different results when using function argmax or argmin for data type ".cuda()" and ".cpu()" enhancement good first issue module: docs todo triaged

#10370 opened Aug 9, 2018 by heylfda

6

Improve docs search engine indexing high priority module: doc infra module: docs triaged

#32838 opened Jan 30, 2020 by josherich

2

Easy way to create previews of the docs website after any changes module: doc infra triaged

#28035 opened Oct 15, 2019 by zou3519

1

[doc] Tensor.mean: dtype kwarg is not documented module: docs triaged

#29758 opened Nov 13, 2019 by SsnL

1

Tutorial Error for "Using the PyTorch C++ Frontend" module: docs triaged

#31037 opened Dec 10, 2019 by lzxzy

2

unsqueeze has 'out=' option documented but not implemented(?) module: docs triaged

#29800 opened Nov 14, 2019 by meganset

1

Implement erfcx, logerfc, logerfcx special functions feature module: operators topic: numerical-stability triaged

#31945 opened Jan 8, 2020 by cossio

4

matmul: no warning when contracting differently named dimensions module: named tensor triaged

#33006 opened Feb 5, 2020 by Stannislav

Dockerfile for people to quick-start contributing enhancement module: docker triaged

#32030 opened Jan 10, 2020 by vadimkantorov

8

Documentation makefile should include torchvision module: doc infra module: docs triaged

#27708 opened Oct 10, 2019 by zou3519

[jit] C++ Documentation jit jit-backlog

#28519 opened Oct 23, 2019 by driazati

1

Add javasphinx to docs/requirements.txt module: docs triaged

#27705 opened Oct 10, 2019 by zou3519

4

torch.logsumexp only works on a single dimension, but doc says tuple of dims should work module: docs triaged

#31084 opened Dec 11, 2019 by chausies

[Feature Request] Make torch.solve output NaN for singular matrix topic: crash triaged

#31546 opened Dec 21, 2019 by ducha-aiki

3

Add instructions for building torch.distributed on macOS module: distributed module: docs triaged

#29202 opened Nov 5, 2019 by pietern

Out-of-date link to pytorch ci dockerfiles in jenkins readme module: docs triaged

#32087 opened Jan 12, 2020 by Enealor

TensorIterator "builder" options should be documented. module: internals triaged

#24264 opened Aug 13, 2019 by gchanan

Exclude generated source docs from Google module: doc infra triaged

#26123 opened Sep 12, 2019 by ezyang

16

Tutorials should be run in CI high priority module: ci module: docs releng triaged

#32004 opened Jan 9, 2020 by ezyang

5

Optional seq_len argument to torch.nn.utils.rnn.pad_sequence enhancement module: operators triaged

#32912 opened Feb 1, 2020 by adamlerer

1

Be able to build torch.distributed documentation easier module: distributed module: doc infra triaged

#27843 opened Oct 14, 2019 by zou3519

7

Cannot select version in the tutorials page module: docs triaged

#28938 opened Oct 30, 2019 by z-a-f

1

TestTorch.test_doc should be in TestDocCoverage module: doc infra module: docs triaged

#28033 opened Oct 15, 2019 by zou3519

6

[feature request] Docs for fuse_conv_bn_eval module: docs triaged

#28757 opened Oct 27, 2019 by vadimkantorov

3

Why doc building isn't failing us for referring to a non-existent method? module: docs triaged

#25991 opened Sep 11, 2019 by Nikronic

2

update docs that sorting is not needed in module: docs topic: rnn triaged

#23079 opened Jul 19, 2019 by brando90

1

Fix type hints for named tensor enhancement module: named tensor module: typing triaged

#27846 opened Oct 14, 2019 by zou3519

[doc] functionalities not documented good first issue module: docs triaged

#9886 opened Jul 26, 2018 by SsnL

10

Tensor.random_ should be able to generate all 64 bit numbers including min and max value module: random triaged

#33299 opened Feb 13, 2020 by pbelevich

1

Redundant counter in batchnorm impl module: nn triaged

#30440 opened Nov 26, 2019 by soumyarooproy

2

[cppdocs] the docs of class Tensor is missing high priority module: cpp module: docs triaged

#25845 opened Sep 9, 2019 by jimages

1

Confusing example of broadcasting in torch.matmul module: docs triaged

#22763 opened Jul 11, 2019 by chmendoza

1

Warn / clearly document that you can't use intermediate results in saved tensors for higher order derivatives actionable module: autograd module: docs triaged

#23108 opened Jul 19, 2019 by ezyang

1

[C++] Add strong wolfe line search to the LibTorch interface enhancement module: cpp triaged

#33284 opened Feb 13, 2020 by chenzhekl

1

Add Wishart and inverse-Wishart distributions feature module: distributions triaged

#29044 opened Nov 1, 2019 by dccastro

9

Title of docs page includes "PyTorch master documentation" even for non-master branches. module: docs triaged

#21290 opened Jun 3, 2019 by umanwizard

1

Add simple tutorial for model parallel training high priority module: docs module: rpc triaged

#29537 opened Nov 11, 2019 by rohan-varma

8

[docs] torch.tensordot example is a little confusing module: docs triaged

#22748 opened Jul 11, 2019 by zou3519

Can't build documentation on master high priority module: doc infra module: docs triaged

#32703 opened Jan 28, 2020 by zou3519

11

IndexExpressions (or slice) for jit.script functions jit triaged

#31545 opened Dec 21, 2019 by narfanar

2

[doc] nn.Module.forward documentation unclear enhancement module: docs module: nn triaged

#22220 opened Jun 25, 2019 by henon

6

Transformer Lack of Embedding Layer and Positional Encodings feature module: docs triaged

#24826 opened Aug 18, 2019 by TrentBrick

6

`TestDocCoverage.test_torch` error messages could be clearer module: docs triaged

#30759 opened Dec 4, 2019 by zou3519

`pad_packed_sequence` outdated documentation module: docs module: nn topic: rnn triaged

#25497 opened Aug 30, 2019 by cai-lw

The guidelines for loading a PyTorch model in C++ do not work on Windows module: docs triaged

#30379 opened Nov 24, 2019 by puhach

1

Mnasnet0_5 first layer shape incorrect module: vision triaged

#31471 opened Dec 19, 2019 by mattthelee

3

make torch.utils._download_url_from_file public and add a docstring module: operators topic: operator triaged

#23305 opened Jul 24, 2019 by soumith

Truncated normal distribution feature module: distributions triaged

#32293 opened Jan 16, 2020 by cossio

9

torch.full_like missing documentation for out input variable enhancement module: docs module: operators triaged

#20935 opened May 24, 2019 by omry

1

ONNX verbose=True does not print onnx initializers module: onnx triaged

#28834 opened Oct 29, 2019 by ezyang

4

Confusing error messages of tensor.scatter_ on both CPU and CUDA module: operators small topic: error checking triaged

#32872 opened Jan 31, 2020 by vadimkantorov

Request for Lint Pass to Detect Modification on Parameters/Attributes during TorchScript Inference feature jit

#31202 opened Dec 12, 2019 by houseroad

1

Reverse Cumulative Sum feature module: operators triaged

#33520 opened Feb 19, 2020 by lematt1991

torch.tensordot has inconsistent signature with torch script jit

#32695 opened Jan 28, 2020 by ferrine

Bug in add_param - functionality for tensorboard (scatter matrix view) module: tensorboard visualization

#32528 opened Jan 23, 2020 by zero-lummel

1

Warning when link libtorch and opencv4.2.0 together better-engineering module: build triaged

#33049 opened Feb 6, 2020 by stereomatchingkiss

3

Make _rpc_sync_torchscript and _rpc_async_torchscript work with autograd profiler jit triaged

#32375 opened Jan 17, 2020 by zhaojuanmao

Add a private API that lists overridable functions enhancement module: tests triaged

#33182 opened Feb 11, 2020 by ngoldbaum

3

[FR] nn.init.* accept None as input tensor enhancement module: nn triaged

#30535 opened Nov 27, 2019 by SsnL

Improve error message in nn.MultiheadAttention module: nn triaged

#28719 opened Oct 25, 2019 by vadimkantorov

2

rpc.shutdown() does not abort pending sends created in separate threads. high priority module: rpc triage review

#33208 opened Feb 11, 2020 by rohan-varma

Tensor.random_(from, to, ...) should check that from/to are in tensor's dtype bounds module: random triaged

#33106 opened Feb 7, 2020 by pbelevich

1

Improve documentation of torch.lu module: docs triaged

#22552 opened Jul 5, 2019 by arvidfm

2

Boolean Index error when it comes to a small 2d tensor module: operators triaged

#29522 opened Nov 10, 2019 by happyalfred2016

2

[C++] Don't use DeprecatedTypeProperties in torch::utils::reorder_tensors_like module: internals triaged

#32427 opened Jan 20, 2020 by Baranowski

torch.sum(tensor, dim=()) is different from np.sum(arr, axis=()) high priority quansight-nack status: red topic: TensorIterator topic: bc-breaking topic: deprecation triaged

#29137 opened Nov 4, 2019 by zou3519

28

Allow explicit gradients in torch.distributed.autograd.backward() API module: rpc triaged

#27517 opened Oct 8, 2019 by pritamdamania87

1

We should suggest using as_strided_ instead of set_ module: docs triaged

#28558 opened Oct 23, 2019 by gchanan

[docs] torch.set_flush_denormal(...) to mention default mode module: docs triaged

#19652 opened Apr 24, 2019 by vadimkantorov

1

[jit] Python type hints in TorchScript classes don't work jit

#31386 opened Dec 17, 2019 by driazati

JIT, nn.utils.weight_norm and {save,load}_state_dict produce wrong results jit

#30754 opened Dec 4, 2019 by IggShaman

4

[FR] More consistent matrix norm for torch.norm high priority module: numpy module: operators topic: bc-breaking topic: linear algebra triaged

#24802 opened Aug 17, 2019 by SsnL

6

Port `masked_fill` operator from the TH code to Aten better-engineering module: cpu module: cuda module: operators topic: porting triaged

#24303 opened Aug 14, 2019 by pbelevich

Better documentation about PyTorch's dependencies module: build module: docs module: third_party triaged

#23768 opened Aug 4, 2019 by bkmgit

1

Implement derivative for torch.lstsq topic: derivatives topic: linear algebra triaged

#27036 opened Sep 29, 2019 by carlosgmartin

1

torch.full is inconsistent with np.full module: numpy module: operators topic: bc-breaking triaged

#27171 opened Oct 1, 2019 by zou3519

3

C++ randint returns float32, python returns int64 module: cpp triaged

#32166 opened Jan 14, 2020 by meganset

3

Remove deprecated codepath for old-style autograd.Function actionable better-engineering fixathon module: autograd topic: deprecation triaged

#30696 opened Dec 3, 2019 by albanD

2

[ONNX] torch.onnx emits -1 in shape for Expand module: onnx triaged

#32926 opened Feb 3, 2020 by shinh

1

[docs] Automatically detect docs missing in rst enhancement module: docs module: tests triaged

#20481 opened May 14, 2019 by vadimkantorov

4

Lint rule to test for creation of tensor in native/ without options() better-engineering module: build module: lint triaged

#26967 opened Sep 27, 2019 by ezyang

[doc] Document general guidelines to work with CUDA async copying and streams module: cuda module: docs triaged

#19571 opened Apr 22, 2019 by SsnL

2

Autogenerate code example / tutorial outputs in documentation good first issue module: doc infra module: docs todo triaged

#6662 opened Apr 17, 2018 by zou3519

15

Split hypothesis tensor to generate different types triaged

#28471 opened Oct 22, 2019 by z-a-f

Remove RPC internal helper that overrides the default pickler module: rpc triaged

#30633 opened Dec 2, 2019 by pritamdamania87

1

torch.quantized.modules.floatFunctional -- add a little color commentary to the doc module: docs quantization_release_1.3 triaged

#26990 opened Sep 27, 2019 by gottbrath

1

C++ / JIT Parity for ops in `torch/functional.py` and `torch/tensor.py` jit

#30786 opened Dec 5, 2019 by eellison

TripletMarginLoss example isn't clear module: docs triaged

#20143 opened May 5, 2019 by PetroSokirniy

2

Add get_type() API to RRef module: rpc triaged

#33210 opened Feb 11, 2020 by rohan-varma

4

logsumexp with subtraction feature triaged

#32097 opened Jan 13, 2020 by fuzic

`clip_grad_norm` allows negative `max_norm` values enhancement module: nn triaged

#31367 opened Dec 17, 2019 by araffin

Use of word "elements" in `torch.utils.data` samplers module: docs triaged

#21779 opened Jun 14, 2019 by pwiercinski

2

last updated timestamp enhancement module: docs triaged

#30403 opened Nov 25, 2019 by vainaijr

[debatable] Better infer dtype in torch.as_tensor enhancement topic: operator triaged

#33040 opened Feb 6, 2020 by vadimkantorov

5

Size checks in recurrent cells should be moved from python to c++ implementations enhancement topic: performance triaged

#32193 opened Jan 14, 2020 by ngimel

Store and retry futures that are created when nodes cleanup DistAutogradContext better-engineering high priority module: rpc triaged

#28123 opened Oct 16, 2019 by rohan-varma

1

"git describe" shows incorrect version 1.0 instead of 1.2 module: build triaged

#26258 opened Sep 15, 2019 by mratsim

2

a trivial question in torch.nn.init._calculate_fan_in_and_fan_out module: nn triaged

#29089 opened Nov 2, 2019 by seanywang0408

2

ProcessGroupMPI reports incorrect world size module: distributed triaged

#28244 opened Oct 17, 2019 by pritamdamania87

2

Move ProcessGroup C++ tests to use gtest better-engineering enhancement module: distributed triaged

#28410 opened Oct 22, 2019 by rohan-varma

1

CI test should use PR commit instead of pulling the latest master module: ci triaged

#31788 opened Jan 2, 2020 by mrshenli

2

Statically checked tensor shapes feature module: internals triaged

#26889 opened Sep 26, 2019 by vsiles

15

Remove observer in qat modules quantization topic: quantize script triaged

#27622 opened Oct 9, 2019 by jerryzh168

2

Default shuffle behavior of DistributedSampler module: dataloader module: distributed triaged

#31232 opened Dec 13, 2019 by viperit

[RPC] Add type annotations for RPC-related Python files better-engineering module: rpc triaged

#24130 opened Aug 9, 2019 by mrshenli

Implement torch.xlogy actionable feature module: operators triaged

#22656 opened Jul 9, 2019 by colesbury

6

Use of Sequence collections for abstract classes in Dataset feature module: dataloader triaged

#31657 opened Dec 27, 2019 by diegovincent

Warn about driver CUDA mismatch in torch.cuda.is_available() enhancement module: cuda triaged

#30528 opened Nov 27, 2019 by colesbury

2

NLLLoss documentation is inscrutable module: docs

#17577 opened Feb 28, 2019 by gchanan

2

Python/C++ API Parity: torch.optim optimizers module: cpp triaged

#28440 opened Oct 22, 2019 by yf225

3

[c10] c10 dispatch doesn't support tracing of scalars high priority module: dispatch module: internals triaged

#27097 opened Sep 30, 2019 by jamesr66a

18

Latex Errors when Compiling documentation to latexpdf module: doc infra module: docs triaged

#20183 opened May 6, 2019 by PowerSnail

3

add method to make tensor constant for debug purposes module: autograd triaged

#30458 opened Nov 26, 2019 by noskill

7

Refactor CircleCI config for version 2.1 better-engineering module: ci triaged

#24025 opened Aug 8, 2019 by ezyang

1

'tensorflow_core._api.v2.io.gfile' has no attribute 'get_filesystem' module: tensorboard needs reproduction triaged

#30966 opened Dec 9, 2019 by develooper1994

12

[feature request] Run examples from docs as tests feature module: docs triaged

#20359 opened May 10, 2019 by vadimkantorov

1

torch.stack: bad shape error message actionable module: operators triaged

#29510 opened Nov 9, 2019 by rohit-gupta

3

[quantization] fix the seed for hypothesis in CI quantization triaged

#28323 opened Oct 18, 2019 by z-a-f

3

Move hypothesis installation to docker better-engineering high priority module: tests topic: flaky-tests triaged

#31136 opened Dec 11, 2019 by zou3519

3

tensor.var_mean variant for existing torch.var_mean (and same for std_mean) enhancement module: operators triaged

#24041 opened Aug 8, 2019 by vadimkantorov

1

`index_select` with multidimensional `index` feature topic: advanced indexing triaged

#30574 opened Nov 29, 2019 by carlosgmartin

Documentation inaccurate for CrossEntropyLoss (and other loss functions) for multidimensional case module: docs module: nn triaged

#19525 opened Apr 19, 2019 by physicsrob

1

Pylint Error `torch.tensor is not callable` module: internals triaged

#24807 opened Aug 17, 2019 by Microsheep

7

Enhance autograd profiler to support detailed profiling of RPC, RRef and distributed autograd module: rpc triaged

#28734 opened Oct 26, 2019 by pritamdamania87

Port TH operators to Aten (umbrella issue) better-engineering module: operators topic: porting triaged

#24507 opened Aug 16, 2019 by VitalyFedyunin

5

Get lr scheduler warning when using amp to train with fp16 module: optimizer topic: dependency bug triaged

#27595 opened Oct 9, 2019 by CoinCheung

5

pairwise_dist eps argument is confusing enhancement module: docs module: operators topic: bc-breaking topic: numerical-stability triaged

#30263 opened Nov 21, 2019 by SsnL

4

Label.dim Enforcement Check in AccuracyOp caffe2 triaged

#22375 opened Jun 29, 2019 by fulltopic

Better documentation / molly-guards around use of multiprocessing with spawn in Jupyter/ipython notebooks enhancement module: docs triaged

#20375 opened May 10, 2019 by ezyang

1

Behavior of F.dropout in eval mode low priority module: docs triaged

#26338 opened Sep 17, 2019 by sidml

2

SummaryWriter doesn't read comment if log_dir precised module: tensorboard triaged

#23771 opened Aug 4, 2019 by Kaalaam

1

[JIT] script doesn't convert dtypes back to torch.dtype from long jit triaged

#24102 opened Aug 9, 2019 by jamesr66a

Confusing error message for Custom Class type mismatch jit jit-backlog

#24229 opened Aug 13, 2019 by dzhulgakov

`attn_mask` in nn.MultiheadAttention is additive module: docs module: nn triaged

#21518 opened Jun 7, 2019 by mttk

4

RPC threads should return an error message instead of crash on failures enhancement module: rpc triaged

#30819 opened Dec 5, 2019 by mrshenli

2

[ONNX] support index_fill enhancement module: onnx triaged

#26138 opened Sep 12, 2019 by dashesy

3

Gloo scatter gives wrong result for stride != 1 module: distributed topic: bootcamp triaged

#24836 opened Aug 19, 2019 by gaohuazuo

3

Slow (20-50x) RNN tutorial/example when torch is installed using pip comp. to conda installation module: binaries module: mkl topic: performance topic: rnn triaged

#29722 opened Nov 13, 2019 by CeadeS

18

Tensorboard logging image WITH LABEL module: tensorboard triaged

#27030 opened Sep 29, 2019 by jinyu121

2

`Tensor.__reversed__` breaks protocol for reversible objects module: internals module: pybind triaged

#28577 opened Oct 24, 2019 by henribru

2

Pin flake8 version in CI module: ci triaged

#24188 opened Aug 12, 2019 by zou3519

torch.utils.tensorboard.SummaryWriter fails to flush at program exit module: tensorboard triaged

#24234 opened Aug 13, 2019 by vadimkantorov

6

[RPC] Fix logging initialization warning in ProcessGroupAgent module: internals triaged

#24956 opened Aug 21, 2019 by xush6528

Add automatic tuning flags to utils.data.dataloader feature low priority module: dataloader triaged

#23048 opened Jul 18, 2019 by OBannon37

3

[ONNX] torch.arange with non-integer input module: onnx triaged

#27718 opened Oct 10, 2019 by dashesy

Negative step in index gives confusing error in JIT jit jit-backlog triaged

#25135 opened Aug 24, 2019 by Chillee

Zero-dim Tensors (scalars) should be printed at full precision module: printing triaged

#21028 opened May 28, 2019 by colesbury

4

Add in-place view (view_) enhancement module: operators triaged

#27276 opened Oct 3, 2019 by BradMcDanel

1

Broken in-line code in documentation module: docs

#17243 opened Feb 19, 2019 by seungwonpark

2

[feature req. for torch.inverse] optional return for indicating that whether input is not invertable or not. enhancement module: operators topic: linear algebra triaged

#25095 opened Aug 23, 2019 by 7starsea

2

[FR] torch.(Generator|random).seed allows specifying the seed value enhancement module: random triaged

#25070 opened Aug 23, 2019 by SsnL

Test `make html-stable` target in CI module: doc infra triaged

#27766 opened Oct 11, 2019 by zou3519

The documentation to Module._version isn't visible. module: docs

#17612 opened Mar 1, 2019 by PiotrCzapla

5

Make MultiProcessTestCase pickable better-engineering module: distributed module: tests triaged

#23849 opened Aug 6, 2019 by xush6528

1

Automatic rank selection when using file:// initialization method enhancement module: distributed triaged

#22128 opened Jun 24, 2019 by pietern

documentation for C++ / libtorch autograd profiler module: cpp module: docs triaged

#18630 opened Mar 29, 2019 by abhiutd

3

Support sublist arguments for torch.einsum enhancement feature low priority module: numpy module: operators triaged

#21412 opened Jun 5, 2019 by LeviViana

6

[C++] `Module::pretty_print` is broken module: cpp module: nn triaged

#25979 opened Sep 11, 2019 by chenzhekl

5

[JIT] Allow user to provide aliasing information on input tensors and model parameters jit triage review

#28191 opened Oct 17, 2019 by jjsjann123

2

Symeig unstable in float precision topic: linear algebra triaged

#24466 opened Aug 16, 2019 by lucmos

1

Confusing documentation with distributions.Categorical about logits module: distributions module: docs triaged

#16291 opened Jan 24, 2019 by W4ngatang

1

Feature request: Fix dimension convention for masks in transformer feature module: nn triaged

#25132 opened Aug 23, 2019 by malmaud

3

Lint rule to prevent direct use of #pragma omp module: lint triaged

#20466 opened May 14, 2019 by ezyang

2

Build reconfiguration should consistently honor env variables module: build triaged

#23512 opened Jul 29, 2019 by xuhdev

2

Make topk sort stable OSS contribution wanted enhancement high priority module: operators quantization topic: determinism triaged

#27542 opened Oct 8, 2019 by volcacius

14

[docs] Missing argument description (value) in scatter_ function documentation module: docs todo

#16014 opened Jan 14, 2019 by jac99

3

cpp_extension should use ninja for setup.py builds too module: build module: cpp-extensions triaged

#23384 opened Jul 25, 2019 by ezyang

1

Rename ignore_index to ignore_target in CrossEntropyLoss enhancement low priority module: nn small triaged

#18206 opened Mar 20, 2019 by nishankjain

3

BCEWithLogitsLoss expects wrong shape of weight (#classes instead of batch size) module: nn topic: loss triaged

#25174 opened Aug 26, 2019 by phillies

1

The number zero (0) looks like a lowercase "o" in some places in the docs module: docs

#14965 opened Dec 9, 2018 by jameshamm

Support `clamp()` with tensor min and max enhancement good first issue topic: bootcamp triaged

#2793 opened Sep 19, 2017 by zuoxingdong

7

[sparse] add descriptions and examples for methods at torch.sparse doc page

#14112 opened Nov 16, 2018 by weiyangfb

1

Wrong description of positive class weight in BCEWithLogitsLoss module: docs triaged

#16804 opened Feb 6, 2019 by siarez

2

Build instructions clarification module: build module: docs small triaged

#25464 opened Aug 30, 2019 by hxtk

2

Documentation is slow and scroll-jumpy module: docs

#15760 opened Jan 5, 2019 by guillaume-chevalier

1

Make it easier to bisect on PyTorch better-engineering module: build module: ci triaged

#22868 opened Jul 15, 2019 by ezyang

2

Difference between dropout2d and dropout3d module: nn triaged

#22879 opened Jul 15, 2019 by kianasun

2

[ONNX] Export boolean indexing using where instead of Gather module: onnx triaged

#26103 opened Sep 12, 2019 by dashesy

1

The speed of `torch.einsum` and `torch.matmul` when using `fp16` is slow module: cuda topic: performance triaged

#23061 opened Jul 19, 2019 by fyubang

12

Large documentation pages take a long time to load enhancement module: docs triaged

#14686 opened Dec 2, 2018 by robertknight

2

inconsistent behavior while multiplying int and bool values module: operators triaged

#24288 opened Aug 13, 2019 by izdeby

1

Generated `__init__.pyi` contains invalid default values todo

#17257 opened Feb 19, 2019 by sproshev

latex of init.norm is wrong enhancement module: docs module: nn triaged

#20903 opened May 24, 2019 by selaselah

1

wrong notations as markdown of jupyter notebooks in tutorials module: docs

#17635 opened Mar 2, 2019 by eduidl

1

Consolidate multiprocessing helpers in distributed tests module: distributed triaged

#23230 opened Jul 23, 2019 by pietern

Document whether it is possible to train TorchScript modules jit module: cpp module: docs triaged

#17614 opened Mar 1, 2019 by martinruenz

8

fractional_max_pool2d_with_indices silently ignores output_ratio if output_size is provided module: operators topic: error checking triaged

#24031 opened Aug 8, 2019 by zou3519

1

CPU torch.exponential_ function may generate 0 which can cause downstream NaN module: random triaged

#22557 opened Jul 5, 2019 by colesbury

[jit] Optional type refinement on non-named expressions jit triaged

#22155 opened Jun 24, 2019 by driazati

2

Consider changing the behavior of Tensor.__contains__(Tensor) to make more sense module: operators triaged

#24338 opened Aug 14, 2019 by zou3519

6

Port dragon4_scientific for pretty float tensor print. todo

#12797 opened Oct 17, 2018 by ailzhang

1

Tensorboard: Add disable flag for debugging enhancement module: tensorboard triaged

#24236 opened Aug 13, 2019 by Haydnspass

2

documentation for adding a new type via C++ extensions module: docs

#15607 opened Dec 29, 2018 by soumith

12

C++ module API footgun: assigning to parameter doesn't update `parameters()` list module: docs triaged

#21682 opened Jun 12, 2019 by ezyang

3

Support argument dim for torch.flip (currently it's only dims) module: operators small triaged

#18095 opened Mar 16, 2019 by vadimkantorov

7

Please document torch.backend.cudnn.version() module: docs

#12468 opened Oct 8, 2018 by jlquinn

Provide better documentation for torch.Size module: docs triaged

#12532 opened Oct 10, 2018 by Rizhiy

6

Provide option to use alias method in Categorical.sample() module: distributions triaged

#18906 opened Apr 4, 2019 by neerajprad

`nn.Linear` allows 1d input tensors triage review

#16896 opened Feb 8, 2019 by jatentaki

1

[jit] In pickler, don't memoize if not necessary jit jit-backlog triaged

#21544 opened Jun 7, 2019 by driazati

torch._C._demangle("t") needs to return `t` instead of `unsigned short` module: autograd module: internals triaged

#19958 opened Apr 30, 2019 by dnlcrl

5

The word "ints" is not rendered correclty in docs. module: docs todo

#16523 opened Jan 29, 2019 by umanwizard

LRN type mismatch when exported to onnx high priority module: onnx triaged

#18281 opened Mar 21, 2019 by lara-hdr

4

Unify tensor shape formatting in shape checks enhancement module: operators topic: convolution topic: error checking triaged

#22919 opened Jul 16, 2019 by vadimkantorov

4

Inconsistant values of lr_scheduler.get_lr and lr in optimizer.param_groups module: optimizer triaged

#20138 opened May 5, 2019 by lihuanglx

2

Remove norm_except_dim (because norm now provides the same functionality) module: operators topic: bc-breaking topic: bootcamp triaged

#17395 opened Feb 22, 2019 by t-vi

11

How to use libtorch api torch::nn::parallel::data_parallel train on multi-gpu module: cpp module: distributed module: docs module: nn topic: multi-gpu topic: performance triaged

#18837 opened Apr 4, 2019 by DDFlyInCode

4

[docs] Script for releasing new versions of the docs

#9894 opened Jul 26, 2018 by zou3519

Caffe2 C++ tutorial is not working caffe2

#14337 opened Nov 23, 2018 by rribani

1

[caffe2] Documentation of Optimizers (Python API) caffe2

#11976 opened Sep 22, 2018 by CarlosYeverino

[Docs] Update broadcasting documentation for scalars / n-dimensional empty tensors module: docs todo

#10377 opened Aug 9, 2018 by gchanan

2

Statically make `__setstate__` set all attributes/parameters module: serialization triaged

#20497 opened May 14, 2019 by driazati

1

Math doesn't render module: docs todo

#14991 opened Dec 10, 2018 by beneyal

1

CMAKE_PARSE_IMPLICIT_LINK_INFO Function invoked with incorrect arguments has workaround module: build topic: flaky-tests triaged

#22682 opened Jul 10, 2019 by ezyang

1

Improve doc page readability

#11968 opened Sep 22, 2018 by Evpok

7

Document what is C10 module: docs triaged

#14850 opened Dec 6, 2018 by christoph-conrads

4

Update weight initialisations to current best practices feature high priority module: nn quansight-nack topic: bc-breaking triaged

#18182 opened Mar 19, 2019 by Kaixhin

8

[JIT] Source highlighting doesn't line up when tabs are used for indentation enhancement jit jit-backlog triaged

#20119 opened May 4, 2019 by jamesr66a

1

as_tensor does not use the device of the default tensor type triage review

#14996 opened Dec 10, 2018 by mtikekar

torch.from_PIL() Request ? feature module: vision triaged

#18998 opened Apr 7, 2019 by MohamedAliRashad

2

Name of "ResetWorkSpace" method in workspace class in docs is wrong ("S")

#12595 opened Oct 12, 2018 by hosjiu1702

torch.Tensor.cpu talks about the object being "on the correct device" module: docs

#17553 opened Feb 27, 2019 by janden

torch.distributions.Binomial.sample() uses a massive amount of memory module: distributions topic: memory usage triaged

#20343 opened May 10, 2019 by jamestwebber

11

[Caffe2] Which document should we follow to install caffe2? caffe2

#10341 opened Aug 8, 2018 by lianglia-apollo

1

[docs] Document leaf node distinction in FAQ or Tutorials module: docs

#10065 opened Jul 31, 2018 by GZHermit

1

worker_init_fn has no examples module: docs

#13555 opened Nov 4, 2018 by SreenivasVRao

2

improve jit error message for legacy constructor module: docs module: onnx triaged

#18434 opened Mar 25, 2019 by soumith

3

torch.jit.script is not getting type hint correct jit low priority

#16492 opened Jan 29, 2019 by zasdfgbnm

1

pytorch_doc_push is racing with itself

#14702 opened Dec 3, 2018 by ezyang

3

[Feature Request] Common constants in the torch.* namespace enhancement module: numpy triaged

#19172 opened Apr 11, 2019 by soumith

5

Improve save() method in torch.jit.ScriptModule jit jit-backlog low priority triaged

#17688 opened Mar 5, 2019 by laughmetal

4

Better dev docs for writing native CPU kernels with Vec256 medium priority (this tag is deprecated)

#10956 opened Aug 28, 2018 by SsnL

tutorial_blob ERROR caffe2

#12808 opened Oct 18, 2018 by VisionZQ

1

caffe2 argmax and argmin documentation incorrect for output type caffe2

#11890 opened Sep 20, 2018 by cdepillabout

[docs] Mention inline CUDA extensions in tutorial module: docs todo

#10253 opened Aug 6, 2018 by vadimkantorov

SyncBatchNorm should support 2D input (B, C) enhancement module: distributed module: nn triaged

#20204 opened May 7, 2019 by sshaoshuai

8

Descriptive error message for distribution constraint violation module: distributions topic: error checking triaged

#18133 opened Mar 18, 2019 by jhostetler

5

Suggest model.eval() in torch.no_grad (and vice versa) module: docs triaged

#19160 opened Apr 11, 2019 by HaleTom

6

The `unary_kernel` call should be pushed into CopyKernel.cpp and it should completely replace the current `copy_kernel`. module: cpu triaged

#19437 opened Apr 18, 2019 by VitalyFedyunin

Better include path when compiling mkldnn medium priority (this tag is deprecated) module: build

#16708 opened Feb 4, 2019 by SsnL

2

No code example for AdaptiveLogSoftmaxWithLoss module: docs triaged

#14318 opened Nov 22, 2018 by beneyal

1

String in tensor module: internals triaged

#20412 opened May 12, 2019 by Alaya-in-Matrix

2

RFC: Add torch.experimental.deterministic flag to force deterministic algorithms feature high priority module: internals topic: determinism triaged

#15359 opened Dec 18, 2018 by colesbury

26

Conv2d layers should accept 4-tuple for padding argument feature

#17150 opened Feb 15, 2019 by sytelus

2

In Tutorials, Using the Pytorch C++ Frontend, return linear(input) + another_bias, should be linear->(input), on Win10

#16555 opened Jan 30, 2019 by fakerbrother

2

[Documentation Clarification] nn.Upsample todo

#7413 opened May 9, 2018 by Naman-ntc

Port multinomial to Aten

#12499 opened Oct 9, 2018 by ailzhang

1

[caffe2] the caffe2 operators document is too old caffe2

#10738 opened Aug 21, 2018 by JerryShih

Misleading step method in lr_scheduler.ReduceLROnPlateau medium priority (this tag is deprecated) todo

#12146 opened Sep 27, 2018 by LeviViana

1

write a better test that checks that cuda is lazy-loaded medium priority (this tag is deprecated)

#13180 opened Oct 26, 2018 by anderspapitto

1

[doc] specify `not allocate new memory` or `share underlying storage` at ops like `select()`

#10493 opened Aug 13, 2018 by weiyangfb

computing entropy of a tensor feature triaged

#15829 opened Jan 8, 2019 by isalirezag

10

ROCm 2.1: test_gamma_gpu_sample test fails

#16661 opened Feb 1, 2019 by eukaryote31

[C++] better slice or advanced indexing support module: cpp

#15613 opened Dec 29, 2018 by Rikorose

3

We're binding a bunch of crap to 'torch' namespace which shouldn't be there

#16322 opened Jan 24, 2019 by ezyang

7

How to retrain the modelzoo model in Caffe2? caffe2

#16436 opened Jan 28, 2019 by PayalNG

DataLoader with option to re-use worker processes feature high priority module: dataloader triaged

#15849 opened Jan 9, 2019 by dashesy

39

torch int times python float is int topic: type promotion triaged

#15056 opened Dec 11, 2018 by petered

2

Dropout on integer tensor types dies with SIGFPE module: nn topic: crash triaged

#15786 opened Jan 7, 2019 by alphadl

5

Code review on .circleci/

#17342 opened Feb 21, 2019 by ezyang

1