Skip to content

Issues: pytorch/ao

[RFC] torchao Contributor Guide
#391 opened Jun 18, 2024 by jerryzh168
Open 16
float8 upcoming feature tracker
#556 opened Jul 30, 2024 by vkuzo
Open 1
Multibackend tracker
#1082 opened Oct 15, 2024 by msaroufim
Open
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

213 Open 173 Closed
213 Open 173 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

torch.compile error when running the HuggingFace torchao example
#1705 opened Feb 12, 2025 by vkuzo
@HDCharles
1
int4_weight_only api got error when saving transformers models
#1704 opened Feb 12, 2025 by jiqing-feng
2
Model size after quantization quantize question Further information is requested
#1701 opened Feb 11, 2025 by TaylorYangX
3
[DOC] Questions on Integrating a New CPU Operator into TorchAO？ cpu question Further information is requested
#1699 opened Feb 11, 2025 by Zijie-Tian
1
[Fp8 Training Feature Request] Smooth SwiGlu and Configurable AdamWFp8
#1691 opened Feb 10, 2025 by vasqu
5
migration of quantize_ workflow configuration from callables to configs
#1690 opened Feb 10, 2025 by vkuzo
@vkuzo
Performance comparison NF4Tensor vs. BNB Params4bit performance
#1686 opened Feb 10, 2025 by psinger
9
FlexAttention?
#1685 opened Feb 10, 2025 by johnnynunez
6
deprecation of delayed and static scaling from torchao.float8
#1680 opened Feb 6, 2025 by vkuzo
@vkuzo
[Feature Request] Add bias support for torchao/experimental ops
#1675 opened Feb 5, 2025 by metascroy
NF4Tensor and DDP question Further information is requested
#1665 opened Feb 5, 2025 by psinger
1
6
Tensor subclass methods for DTensor and FSDP2 question Further information is requested
#1664 opened Feb 5, 2025 by jeromeku
3
[Needs more investigation] int8_weight_only via quantize_() API on torch.float16 models results in NaN values across multiple CPU architectures bug Something isn't working quantize
#1662 opened Feb 4, 2025 by vmpuri
@leslie-fang-intel
3
Ensure runners / tests are utilizing Torch 2.6
#1655 opened Feb 3, 2025 by drisspg
[Doc] gemlite version question Further information is requested topic: documentation Use this tag if this PR adds or improves documentation
#1653 opened Feb 3, 2025 by bhack
1
make smoothquant more PT2 friendly
#1639 opened Jan 29, 2025 by vkuzo
1
FP8 QAT / FP8 block-wise quantization
#1632 opened Jan 28, 2025 by cassanof
5
Unittests Migration Progress good first issue Good for newcomers
#1621 opened Jan 26, 2025 by osbm
24 of 74 tasks
Wrong result and no speedup with SemiSparseLinear from Torchao compared to torch.nn.Linear
#1617 opened Jan 24, 2025 by lin-ht
3
Performance regression on latest pytorch nightly when using float8_dynamic_activation_float8_weight with granularity == PerTensor
#1609 opened Jan 23, 2025 by vgoklani
2
int4_weight_only Slows Down torch.nn.Linear for Llama2 7B Shapes high priority performance triage review
#1606 opened Jan 23, 2025 by mostafaelhoushi
13
Float8 with FSDP and DelayedScaling: 'WeightWithDelayedFloat8CastTensor' object has no attribute '_tensor'. float8
#1605 opened Jan 23, 2025 by fmo-mt
3
[float8] Add support for blockwise fp8 quantization scheme used in DeepSeek v3 float8 inference topic: new feature Use this tag if this PR adds a new feature
#1594 opened Jan 22, 2025 by danielvegamyhre
3
Sparsity training together with FP8 linear layer in torchao enhancement New feature or request sparsity
#1574 opened Jan 17, 2025 by goldhuang
2
Sam2 video triaged
#1564 opened Jan 14, 2025 by bhack
@cpuhrsch
1
ProTip! Find all open issues with in progress development work with linked:pr.