Search results

TorchAO Missing 3.13T (free-threading) Wheels

Status: Open.

#3243 In pytorch/ao;

· Qubitium opened on Oct 24, 2025

Need Windows build in CI

Status: Open.

#3237 In pytorch/ao;

· Xia-Weiwen opened on Oct 24, 2025

Quantized OpenVLA 7B (4-bit) runs slower than default model

Status: Open.

#3236 In pytorch/ao;

· arunmadhusud opened on Oct 24, 2025

nvfp4: QuantizedNVFP4Kwargs not serializeable

Status: Open.

#3235 In pytorch/ao;

· vkuzo opened on Oct 23, 2025

nvfp4: doesn't work with vLLM + lm-eval

Status: Open.

#3234 In pytorch/ao;

· vkuzo opened on Oct 23, 2025

nvfp4: add padding to handle tensor parallel, size might not be divisible by 16

Status: Open.

#3233 In pytorch/ao;

· vkuzo opened on Oct 23, 2025

nvfp4: why do we need to call weight.contiguous for Qwen3 during lm-eval?

Status: Open.

#3232 In pytorch/ao;

· vkuzo opened on Oct 23, 2025

Deprecation for <code>_default</code> in FqnToConfig

Status: Open.

#3229 In pytorch/ao;

· jcaip opened on Oct 22, 2025

question of blockwise quant fp8 training

float8

moe

Status: Open.

Feature

#3226 In pytorch/ao;

· rakkit opened on Oct 22, 2025

[RFC][Feature Proposal] Dynamic 2:4 semi-structured sparsity with sign-flip pruning and gradient-based regrowth

Status: Open.

#3222 In pytorch/ao;

· catswe opened on Oct 21, 2025

building torchao in a fresh conda env does not work

Status: Open.

#3180 In pytorch/ao;

· vkuzo opened on Oct 15, 2025

[Windows][build]it will meet build failures when build with pytorch xpu version on windows

Status: Open.

#3175 In pytorch/ao;

· wincent8 opened on Oct 15, 2025