pytorch/pytorch

Timeout for process termination upon receiving sigterm should be a user configurable parameter 1 day ago

oncall: distributed good first issue triaged bot-triaged oncall: distributed infra ptd-bot-triaged

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Add TORCH_SHOW_CPP_STACKTRACES when TORCH_DISTRIBUTED_DEBUG = detail 4 days ago

oncall: distributed module: bootcamp good first issue triaged enhancement pt_distributed_rampup module: c10d module: distributed_tool bot-triaged oncall: distributed infra ptd-bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

bmm, topk, cholesky, linalg.norm, max with out variants set causing recompilations in torch.compile 6 days ago

good first issue triaged oncall: pt2 module: dynamic shapes PT2-Bug-Bash

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Make tlparse able to show a summary of distinct graph breaks 6 days ago

good first issue triaged oncall: pt2 module: dynamo dynamo-triage-jan2025 module: compile ux

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

foreach_map enhancements 6 days ago

feature good first issue triaged oncall: pt2 module: inductor module: dynamo internal ramp-up task

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Standardize slot handler naming in `torch/_dynamo` 8 days ago

good first issue triaged enhancement oncall: pt2 module: dynamo bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Enable more flake8-bugbear lints 10 days ago

good first issue module: lint triaged enhancement better-engineering

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[Elastic] etcd find_free_port masks socket creation failures 12 days ago

good first issue triaged module: elastic oncall: distributed infra

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[Distributed Tools] MemoryTracker load does not restore operation count 13 days ago

good first issue triaged module: distributed_tool bot-triaged oncall: distributed infra

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[Distributed] Remove duplicate unreachable return in functional collectives fallback 13 days ago

good first issue triaged module: c10d bot-triaged oncall: distributed infra

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[Flight Recorder] read_dir permanently disables cyclic garbage collection 13 days ago

good first issue triaged module: distributed_tool bot-triaged oncall: distributed infra

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[Elastic] FileStore rendezvous leaks the mkstemp file descriptor 14 days ago

good first issue triaged module: elastic bot-triaged oncall: distributed infra

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[Flight Recorder] Replace raw eval when parsing process-group ranks 14 days ago

good first issue triaged module: distributed_tool bot-triaged oncall: distributed infra

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[export] Schematize nn_module_stack serialization 14 days ago

good first issue export-triaged oncall: export

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Can torchrun have a shell completion? 14 days ago

oncall: distributed feature good first issue triaged bot-triaged oncall: distributed infra ptd-bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

extend torch.distributed's `--tee` to log the nodename 15 days ago

oncall: distributed good first issue triaged enhancement module: elastic oncall: r2p bot-triaged oncall: distributed infra ptd-bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Add a test that test/cpython/v3_13/*.diff files stay in sync with the adapted test files 15 days ago

good first issue module: tests triaged enhancement oncall: pt2 module: dynamo bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Add clearer docs for `torch.distributed.<OP>` vs. `torch.distributed.nn.<OP> 18 days ago

oncall: distributed module: docs good first issue triaged enhancement bot-triaged oncall: distributed infra ptd-bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Collective operations do not work with `torch.BoolTensor`s on `gloo` and raise `Invalid scalar type` 18 days ago

oncall: distributed good first issue triaged bot-triaged oncall: distributed infra ptd-bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Document ProcessGroupOptions 19 days ago

triage review oncall: distributed module: docs good first issue triaged bot-triaged oncall: distributed infra ptd-bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

dynamo remap collective improvement 19 days ago

oncall: distributed good first issue triaged enhancement bot-triaged oncall: distributed infra ptd-bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Unify naming in torch/_dynamo/variables/object_protocol.py 20 days ago

good first issue triaged enhancement oncall: pt2 module: dynamo bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

BuiltinVariable.call_next diverges from CPython builtin_next semantics 21 days ago

triage review good first issue module: correctness (silent) oncall: pt2 module: dynamo bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

all_reduce backward only supports 'sum' reduction, got 'avg' 22 days ago

oncall: distributed good first issue triaged bot-triaged oncall: distributed infra

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[Bug Report] inconsistent global src and local rank in dist.nn.functional.broadcast 24 days ago

triage review oncall: distributed good first issue triaged bot-triaged oncall: distributed infra ptd-bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Cannot initialise process group on CPU device with GLOO backend 24 days ago

oncall: distributed good first issue triaged bot-triaged oncall: distributed infra ptd-bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Add docs for torchrun numa binding on website 25 days ago

oncall: distributed module: docs good first issue triaged bot-triaged oncall: distributed infra ptd-bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Passing string to devices in register_backend causes incorrect behavior 26 days ago

oncall: distributed good first issue triaged bot-triaged oncall: distributed infra ptd-bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[RFC] Add process group revoke in torch.distributed 26 days ago

oncall: distributed good first issue triaged bot-triaged oncall: distributed infra

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

My_rank in the implementation of torch.distributed.scatter_object_list should be global 26 days ago

oncall: distributed good first issue triaged module: c10d bot-triaged oncall: distributed infra ptd-bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

LocalTensorMode: all_to_all_single crashes with uneven split sizes 26 days ago

oncall: distributed good first issue bot-triaged oncall: distributed infra ptd-bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

c10d all_gather aborts with Signal 8 (SIGFPE) when tensor.numel() == 0 26 days ago

oncall: distributed good first issue module: c10d bot-triaged oncall: distributed infra ptd-bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Incorrect results for `torch.distributed.gather` for tensor created from permuted NumPy array 26 days ago

triage review oncall: distributed good first issue triaged module: c10d bot-triaged oncall: distributed infra ptd-bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

torch.distributed.new_group() should consolidate the type of rank list 26 days ago

oncall: distributed good first issue triaged module: c10d bot-triaged oncall: distributed infra ptd-bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[Feature request] `torch.export` .save/.load could support `safetensors` and/or `weights_only=True` 26 days ago

good first issue oncall: pt2 export-triaged oncall: export

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[BUG][PyTorch 2.0 Export][quant]:get_source_partitions() may return different matches with same input graph about 1 month ago

oncall: quantization good first issue oncall: pt2 oncall: export

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Improve the error message in `linalg.eigh` when using large matrices about 1 month ago

good first issue triaged module: 64-bit module: mkl module: linear algebra

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

refactor TracingContext to take a more limited subset of ViewAndMutationMeta about 1 month ago

good first issue triaged better-engineering oncall: pt2 module: aotdispatch module: pt2-dispatcher

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

switch more test cases to use MultithreadTestCase about 1 month ago

good first issue triaged module: dtensor

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

skipIfTorchInductor Tracking Issue about 1 month ago

good first issue triaged oncall: pt2 module: inductor

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[doc] functionalities not documented about 1 month ago

module: docs good first issue triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Use typing_extensions.TypeAliasType for better reexport of `__module__` about 1 month ago

good first issue module: typing module: lint triaged actionable

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[PT2] [Hardening] Track recompiles alongside graph breaks in our actual/expected comparison CI runs about 1 month ago

feature good first issue triaged oncall: pt2 module: dynamo

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Enable ruff rule PLW1510 codebase wide about 1 month ago

good first issue module: lint triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Inductor: delete code that extracts out sizevars by inspecting tensor inputs to find a size that handled it about 2 months ago

good first issue triaged better-engineering module: inductor

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Docs are little bit outdated for torch logs about 2 months ago

good first issue module: logging triaged better-engineering oncall: pt2

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Command to reproduce error is incorrect about 2 months ago

needs reproduction good first issue module: tests triaged module: infra module: testing

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[dynamo] [dynamic] `torch.combinations` throws RuntimeError when `(backend=aot_eager, dynamic=True)` 2 months ago

good first issue triaged oncall: pt2 module: dynamic shapes PT2-Bug-Bash

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

`torch.autograd.functional.jacobian` raises `NotImplementedError` at execution time for `strategy='forward-mode'` with `vectorize=False` instead of raising `ValueError` at argument validation 3 months ago

module: autograd good first issue module: error checking triaged enhancement small module: forward ad bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Improve error message for wrong number of arguments in CachingAutotuner 3 months ago

good first issue module: error checking triaged better-engineering actionable ciflow/trunk oncall: pt2 module: inductor module: compile ux

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[BE] Deduplicate auto_functionalized and triton_kernel_wrapper_functional 3 months ago

good first issue triaged actionable oncall: pt2 module: pt2-dispatcher module: user triton

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Obscure error: Expected a value of type 'List[int]' for argument 'sizes' but instead found type 'immutable_list' 3 months ago

good first issue triaged oncall: pt2 module: dynamic shapes module: pt2-dispatcher llm-amenable PT2-Bug-Bash

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[Docathon] Convert `nn.init.rst` from rST to MyST Markdown 3 months ago

module: docs good first issue triaged enhancement topic: docs topic: not user facing easy bot-triaged docathon-2026

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[Docathon] Convert `elastic/run.rst` from rST to MyST Markdown 3 months ago

module: docs good first issue triaged topic: docs topic: not user facing bot-triaged docathon-2026

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Using an empty tensor and `torch.int64` or `torch.bool` for `dtype` of `nanmean()` works while a non-empty tensor doesn't work 4 months ago

good first issue module: error checking triaged module: boolean tensor

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

TopK implementation slower than a custom divide and conquer implementation 4 months ago

module: performance module: cuda good first issue module: cpu triaged module: sorting and selection

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

torch.eye(d) is slow and hogs cpu for d >= 182 4 months ago

module: performance good first issue module: cpu triaged module: multithreading

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

aten.grid_sampler_3d.default is missing a c-shim implementation, using proxy executor as fallback 4 months ago

good first issue triaged oncall: pt2 module: inductor oncall: export module: aotinductor

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[Inductor] Remove fp8 special handling in `triton_utils.py` because Triton supports PyTorch fp8 dtypes 4 months ago

good first issue triaged enhancement small actionable oncall: pt2 module: inductor module: floatx (formerly float8) bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Feature Request: Deterministic Algorithm for MaxPool3d 4 months ago

feature module: nn good first issue triaged module: determinism module: pooling

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[CUDA] Wrong `owidth_for_shape_check` in `AveragePool3d.cu` 4 months ago

module: cuda good first issue triaged module: pooling bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

`scaled_dot_product_attention` docs describe incorrect `attn_mask` shape 5 months ago

module: docs good first issue triaged small module: sdpa bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Add vmap support for torch.linalg.vander 5 months ago

good first issue triaged module: functorch

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

export: `float8_e8m0fnu` fails to save 5 months ago

good first issue oncall: pt2 module: floatx (formerly float8) oncall: export bot-triaged PT2-Bug-Bash

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[dynamo] Missing support for many trivial builtin functions in operator 5 months ago

good first issue triaged oncall: pt2 module: dynamo

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[Dynamo] Support tracing through _get_current_dispatch_mode_stack 5 months ago

good first issue triaged oncall: pt2 module: dynamo dynamo-tensor-subclasses

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[ONNX][Dynamo] TypeError: torchvision_roi_align() takes from 3 to 6 positional arguments but 7 were given in torch 2.12 nightly 5 months ago

module: onnx good first issue triaged OSS contribution wanted bot-triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

torch.compile does not support Python try/except 5 months ago

good first issue triaged oncall: pt2 module: dynamo dynamo-triage-dec2025

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

torch.compile regression 2.9.1: Triton kernel name '_Runner__kernel_name_0' is not defined 6 months ago

good first issue triaged actionable oncall: pt2 module: inductor module: user triton

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[dynamo, BE] get rid of IncorrectUsage exception 6 months ago

good first issue triaged better-engineering oncall: pt2 module: dynamo

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Assignment target is transposed when using jit.script and avanced indexing 6 months ago

oncall: jit good first issue OSS contribution wanted days

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Device check missing in torch.linalg.solve_triangular leading to hard crash 6 months ago

module: crash good first issue triaged module: linear algebra module: mps

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Enable more flake8-pyi ruff checks 6 months ago

good first issue module: lint triaged actionable

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

get_ema_multi_avg_fn() equation is a little confused 7 months ago

module: docs module: optimizer good first issue triaged easy

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[auto functionalize][partitioner] ones_like is not getting recomputed 7 months ago

good first issue oncall: export module: aotinductor

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Cannot Convert Pytorch model with fft_rfftn layers to ONNX using latest torch.onnx.dynamo_export 7 months ago

module: onnx good first issue triaged OSS contribution wanted

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Better error message when trying to run fp16 weights on CPU 7 months ago

good first issue module: error checking triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

test/distributed/_tools/test_sac_ilp.py::TestSACILP::test_sac_ilp_case1 is seemingly not run in CI 7 months ago

oncall: distributed good first issue

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Compilation events should emit profiler events too 7 months ago

good first issue triaged oncall: pt2

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Eradicate hardcoded `/tmp/` path in PyTorch codebase 8 months ago

good first issue triaged better-engineering actionable

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Remove tensor variable default method fallthrough 8 months ago

good first issue triaged enhancement oncall: pt2 module: dynamo

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

LoweringException does not report associated user stack for code that failed 8 months ago

good first issue module: logging triaged oncall: pt2 module: inductor

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Remove unnecessary `ConstantVariable` wrapping in `raise_observed_exception` 8 months ago

good first issue triaged oncall: pt2 module: dynamo

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Report WHY a symbol was created dynamically in symbolic_shapes logs 9 months ago

good first issue triaged oncall: pt2 module: dynamic shapes

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Save a human readable string of type name in check type id guards for debugging 9 months ago

good first issue triaged oncall: pt2 module: dynamo

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

AOTAutograd: functionalization should emit foreach_copy_() instead of copy_() when possible 9 months ago

good first issue triaged module: mta module: functionalization oncall: pt2 module: aotdispatch module: pt2-dispatcher internal ramp-up task

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[inductor] nan_asserts doesn't work for FP8, "RuntimeError: "isinf" not implemented for 'Float8_e4m3fn'" 9 months ago

good first issue triaged module: inductor module: floatx (formerly float8)

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Dynamo infers different return type vs. eager for `torch.ops.aten` 9 months ago

good first issue triaged oncall: pt2 module: decompositions

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Dump bytecode of resumption frames in tlparse 9 months ago

good first issue module: logging triaged oncall: pt2 module: dynamo

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Autogenerate code example / tutorial outputs in documentation 10 months ago

todo module: docs good first issue triaged module: doc infra

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

dynamo (re)compilation issues: shape (1,1), nn.Parameter, mark_dynamic 10 months ago

good first issue triaged oncall: pt2 module: dynamo recompilations

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

`torch.export.export()` fails on GPU with LSTM model: "Cannot access data pointer of Tensor" 10 months ago

good first issue oncall: pt2 export-triaged oncall: export

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

sdp_kernel causes dynamo error on torch.compile(model, fullgraph=True) 10 months ago

good first issue triaged oncall: pt2 module: dynamo dynamo-must-fix module: sdpa

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Dead link in `torch.compile` docs 11 months ago

module: docs good first issue triaged

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Make torch.random.get_rng_state take a device argument 11 months ago

good first issue triaged actionable module: python frontend

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

torch.fx.Tracer.record_stack_traces is broken in torch 2.4.0 11 months ago

good first issue oncall: export

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

torch.nn.InstanceNorm2d throws "mixed dtype" error with track_running_stats set to True 11 months ago

good first issue module: cpu triaged module: norms and normalization

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

If mark_dynamic fails (Not all values of RelaxedUnspecConstraint are valid) due to specialization, error message should print where specialization came from 11 months ago

good first issue triaged oncall: pt2 module: dynamic shapes

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

torch.export fails for expand_copy when implicit=True 11 months ago

good first issue triaged oncall: pt2 oncall: export

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

[Inductor] Inference failed with the compiled model with aminmax operator 12 months ago

module: autograd good first issue triaged oncall: pt2

pytorch/pytorch

101.0K

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Python

#autograd#deep-learning#gpu#machine-learning#neural-network#numpy#python#tensor

Open Issues Need Help