Search results

there should be an easy way to check that N torchao configs are equivalent

Status: Open.

#3062 In pytorch/ao;

· vkuzo opened on Sep 25, 2025

Allow <code>convert_pt2e</code> to work with newer torchao ops

Status: Open.

#3058 In pytorch/ao;

· u-simha opened on Sep 24, 2025

Add support for quantizing non-<code>weight</code> nn.Parameter

Status: Open.

#3047 In pytorch/ao;

· jcaip opened on Sep 23, 2025

make it easier to debug tensor subclass metadata mismatch

Status: Open.

#3046 In pytorch/ao;

· vkuzo opened on Sep 23, 2025

make subclass stack traces more informative by not using <code>_</code> as function name for op overrides

Status: Open.

#3045 In pytorch/ao;

· vkuzo opened on Sep 23, 2025

[RFC] A Proposal to Support More Accelerators for TorchAO

triage review

Status: Open.

#3044 In pytorch/ao;

· orangeH25 opened on Sep 23, 2025

IntWeightonly quantized model slower than default model ( x86 machine, A100)

quantize_quantize_ APIquantize_ API

triaged

Status: Open.

#3040 In pytorch/ao;

· Rakshith12-pixel opened on Sep 22, 2025

Making sure all safeglobal are added when we import torchao

Status: Open.

#3023 In pytorch/ao;

· jerryzh168 opened on Sep 17, 2025

How to use FP8 training with MoE models?

moe

Status: Open.

#3020 In pytorch/ao;

· BIGBALLON opened on Sep 17, 2025

Filling some Int4 tensor feature gaps after tensor subclass migration

good first issueGood for newcomersGood for newcomers

tracker

Status: Open.

#3013 In pytorch/ao;

· jerryzh168 opened on Sep 16, 2025

Add more supported base_configs to QATConfig

qat

Status: Open.

#2992 In pytorch/ao;

· andrewor14 opened on Sep 12, 2025

Quantized model is slower than original model!

Status: Open.

#2989 In pytorch/ao;

· timpiperseek opened on Sep 12, 2025