Search results

Proposal to add a beginner-friendly introduction tutorial for TorchAO

Status: Open.

#3274 In pytorch/ao;

· smishra8 opened on Nov 1, 2025

[DOCS] Quick Start Guide PT2E Example does not work as is. Undefined objects

Status: Open.

#3270 In pytorch/ao;

· cjm715 opened on Oct 31, 2025

[RFC] NVFP4 Rounding Modes

Status: Open.

#3264 In pytorch/ao;

· syed-ahmed opened on Oct 30, 2025

Add support for parameter quantization to all configs

tracker

Status: Open.

#3252 In pytorch/ao;

· jcaip opened on Oct 28, 2025

TorchAO Missing 3.13T (free-threading) Wheels

Status: Open.

#3243 In pytorch/ao;

· Qubitium 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: 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

building torchao in a fresh conda env does not work

Status: Open.

#3180 In pytorch/ao;

· vkuzo opened on Oct 15, 2025