Search results

float8 training + HF transformer model + torch.compile + DDP does not work

Status: Open.

#2586 In pytorch/ao;

· vkuzo opened on Jul 23, 2025

[Feature Request] TorchAO nightly builds for AArch64

Status: Open.

#2585 In pytorch/ao;

· nSircombe opened on Jul 23, 2025

Cannot use LR Schedulers with param groups on AdamW

Status: Open.

#2574 In pytorch/ao;

· Disty0 opened on Jul 20, 2025

Add static quant flow to NVFP4

Status: Open.

#2572 In pytorch/ao;

· drisspg opened on Jul 18, 2025

Int8DynActInt4WeightQuantizer quantize() not propagating the scales_precision argument

Status: Open.

#2571 In pytorch/ao;

· HaroldBenoit opened on Jul 18, 2025

FP8 PerRow quantization (CUDA capability>=9.0)

Status: Open.

#2566 In pytorch/ao;

· zzlin-0629 opened on Jul 17, 2025

Longer term plan for prototype/inductor

Status: Open.

#2555 In pytorch/ao;

· jerryzh168 opened on Jul 15, 2025

[Feature] Wanda++ for LM head pruning

sparsity

Status: Open.

#2517 In pytorch/ao;

· namgyu-youn opened on Jul 10, 2025

match float8 quant primitives

Status: Open.

#2511 In pytorch/ao;

· jerryzh168 opened on Jul 9, 2025

[ Bug Report ] single_param_adam function does amsgrad but maximum is not implemented and amsgrad setted False deafult

Status: Open.

#2497 In pytorch/ao;

· Musa-Sina-Ertugrul opened on Jul 6, 2025

[Feature Req] Can you add *args and **kwargs to improve extensibility ?

triaged

Status: Open.

#2496 In pytorch/ao;

· Musa-Sina-Ertugrul opened on Jul 6, 2025

MPS experimental test is broken in torchao

ci

Status: Open.

#2487 In pytorch/ao;

· metascroy opened on Jul 3, 2025