Search results

4/x: mx cleanup: use kernel_preference instead of gemm_kernel_choice

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

topic: bc-breakingUse this tag if this PR breaks backward compatibilityUse this tag if this PR breaks backward compatibility

topic: improvementUse this tag if this PR is an improvement (doesn't fit into any of the other categories)Use this tag if this PR is an improvement (doesn't fit into any of the other categories)

Status: Open (in progress).

pytorch/aonumber 3385

#3385 In pytorch/ao;

· vkuzo opened on Nov 24, 2025

[sparse] Migrate Float8SemiSparseTensor off of AQT

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

topic: improvementUse this tag if this PR is an improvement (doesn't fit into any of the other categories)Use this tag if this PR is an improvement (doesn't fit into any of the other categories)

Status: Open (in progress).

pytorch/aonumber 3361

#3361 In pytorch/ao;

· jcaip opened on Nov 20, 2025

Unify llama generate utils

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

topic: improvementUse this tag if this PR is an improvement (doesn't fit into any of the other categories)Use this tag if this PR is an improvement (doesn't fit into any of the other categories)

Status: Open (in progress).

pytorch/aonumber 3318

#3318 In pytorch/ao;

· guangyey opened on Nov 10, 2025

[Reland] Remove config functions like <code>int4_weight_only</code> (#3145)

ci-no-td

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

topic: improvementUse this tag if this PR is an improvement (doesn't fit into any of the other categories)Use this tag if this PR is an improvement (doesn't fit into any of the other categories)

Status: Open (in progress).

pytorch/aonumber 3308

#3308 In pytorch/ao;

· andrewor14 opened on Nov 7, 2025

Support mx_tensor and enable it's test on Intel GPU

ciflow/xpulabel used to trigger xpu CI jobslabel used to trigger xpu CI jobs

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

topic: improvementUse this tag if this PR is an improvement (doesn't fit into any of the other categories)Use this tag if this PR is an improvement (doesn't fit into any of the other categories)

xpuIntel XPU related featuresIntel XPU related features

Status: Open (in progress).

pytorch/aonumber 3248

#3248 In pytorch/ao;

· xiaowangintel opened on Oct 27, 2025

[Intel GPU] enable use of dinov2 models for offload benchmark_low_bit_adam

ciflow/xpulabel used to trigger xpu CI jobslabel used to trigger xpu CI jobs

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

topic: improvementUse this tag if this PR is an improvement (doesn't fit into any of the other categories)Use this tag if this PR is an improvement (doesn't fit into any of the other categories)

Status: Open (in progress).

pytorch/aonumber 3191

#3191 In pytorch/ao;

· arlesniak opened on Oct 16, 2025

[Intel GPU] Enable optim SR test

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

topic: improvementUse this tag if this PR is an improvement (doesn't fit into any of the other categories)Use this tag if this PR is an improvement (doesn't fit into any of the other categories)

Status: Open (in progress).

pytorch/aonumber 3055

#3055 In pytorch/ao;

· arlesniak opened on Sep 24, 2025

Fix Llama4 example

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

topic: improvementUse this tag if this PR is an improvement (doesn't fit into any of the other categories)Use this tag if this PR is an improvement (doesn't fit into any of the other categories)

Status: Open (in progress).

pytorch/aonumber 2846

#2846 In pytorch/ao;

· yiliu30 opened on Aug 22, 2025

prototype float8 training quantizeable mm module

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

topic: improvementUse this tag if this PR is an improvement (doesn't fit into any of the other categories)Use this tag if this PR is an improvement (doesn't fit into any of the other categories)

Status: Open (in progress).

pytorch/aonumber 2807

#2807 In pytorch/ao;

· vkuzo opened on Aug 19, 2025

Enable MOE int4 quant on XPU

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

enhancementNew feature or requestNew feature or request

topic: improvementUse this tag if this PR is an improvement (doesn't fit into any of the other categories)Use this tag if this PR is an improvement (doesn't fit into any of the other categories)

Status: Open (in progress).

pytorch/aonumber 2758

#2758 In pytorch/ao;

· liangan1 opened on Aug 13, 2025

Enables the per_tensor lowering patterns for weight per_packing

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

topic: improvementUse this tag if this PR is an improvement (doesn't fit into any of the other categories)Use this tag if this PR is an improvement (doesn't fit into any of the other categories)

Status: Open (in progress).

pytorch/aonumber 2391

#2391 In pytorch/ao;

· choudhary-devang opened on Jun 17, 2025

Add pct_achievable_gemm_tops and pct_achievable_mem_bw to fp8 roofline utils

benchmark

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

topic: improvementUse this tag if this PR is an improvement (doesn't fit into any of the other categories)Use this tag if this PR is an improvement (doesn't fit into any of the other categories)

Status: Open (in progress).

pytorch/aonumber 2120

#2120 In pytorch/ao;

· mreso opened on Apr 23, 2025