Search results

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

torchao installation should be clearly called out in the README.md

Status: Open.

#2483 In pytorch/ao;

· vkuzo opened on Jul 3, 2025

saving tensor subclass state dict to disk and loading it back is broken

Status: Open.

#2478 In pytorch/ao;

· vkuzo opened on Jul 2, 2025

Run float8 test_everything.sh in CI

ci

float8

Status: Open.

#2477 In pytorch/ao;

· danielvegamyhre opened on Jul 2, 2025

add mx inference APIs to mx_formats readme

Status: Open.

#2459 In pytorch/ao;

· vkuzo opened on Jun 30, 2025

How to not decompose the choose_qparams_affine call_func

Status: Open.

#2456 In pytorch/ao;

· lanluo-nvidia opened on Jun 27, 2025

<code>fill_()</code> not implemented for <code>torchao.quantization.linear_activation_quantized_tensor.LinearActivationQuantizedTensor</code>

Status: Open.

#2454 In pytorch/ao;

· sayakpaul opened on Jun 27, 2025

[MoE training] crash with FSDP if shared_expert uses float8 in torchtitan llama4

triaged

Status: Open.

#2453 In pytorch/ao;

· danielvegamyhre opened on Jun 27, 2025

Gradient Checkpoint makes FP8 Training Slow

float8

Status: Open.

#2445 In pytorch/ao;

· Yzx835 opened on Jun 26, 2025