Pull Requests microsoft/onnxruntime

Collapse A10 pools to a single pool

#32691 opened 2026-09-18 16:27 by eserscor

[WebGPU] Add 3D ConvTranspose support

#32688 opened 2026-09-18 08:33 by gyagp

Add weightless hyper-connection fusion operators

#32687 opened 2026-09-18 07:26 by Copilot

[WebGPU] Upgrade Dawn to the latest

#32686 opened 2026-09-18 02:35 by mei1127

Handle null plugin EP allocators

#32683 opened 2026-09-17 20:29 by crvineeth97

Fix OOB read in InferenceContextImpl::getInputData

#32681 opened 2026-09-17 17:47 by bbernhar

Investigate EmbeddingGemma fp16 CPU regression on Linux ARM64

#32680 opened 2026-09-17 17:36 by Copilot

Fix out-of-bounds read in ConvTransposeWithDynamicPads

#32679 opened 2026-09-17 16:26 by bbernhar

Migrate more pipelines to single windows pool

#32678 opened 2026-09-17 16:23 by eserscor

Add support to dump onnx subgraph

#32675 opened 2026-09-17 13:41 by jatinwadhwa921

Optimize CUDA DynamicSparseAttention for large-context inference

#32671 opened 2026-09-17 09:47 by Copilot

KleidiAI SVE2p1 halfgemm kernel integrated

#32670 opened 2026-09-17 09:20 by melkap01-Arm

[DON'T BE REVIEWED]ClearBuffer in advance to Support concurrent use with explicit Session streams

#32669 opened 2026-09-17 08:59 by xiaofeihan1

Fix CUDA GatherBlockQuantized block indexing

#32665 opened 2026-09-17 03:49 by Copilot

Limit AOT function expansion

#32662 opened 2026-09-16 23:23 by apsonawane

Fix: LayerNormFusion should reject shape-expanding scale and bias patterns

#32660 opened 2026-09-16 22:14 by SIDDARTHAREDDY8

Validate packed GQA head counts during shape inference

#32658 opened 2026-09-16 21:08 by danielsongmicrosoft

2, 3, 5, 6 bit quantization exploration

#32657 opened 2026-09-16 20:45 by jiafatom

Optimize CUDA QSA sparse attention indexing

#32656 opened 2026-09-16 20:20 by Copilot

Validate dynamic pads input access

#32654 opened 2026-09-16 20:01 by apsonawane

Validate initializers before shape propagation

#32653 opened 2026-09-16 19:30 by apsonawane

Validate GQA total sequence length initializer

#32651 opened 2026-09-16 19:11 by apsonawane

Validate QDQ axes against transpose permutations

#32649 opened 2026-09-16 18:48 by apsonawane

Handle unknown weight shapes in QDQ optimizer

#32648 opened 2026-09-16 18:30 by apsonawane

Validate initializer types for transpose optimizer

#32647 opened 2026-09-16 18:20 by apsonawane

Optimize WebGPU GatedDeltaNet recurrent execution

#32645 opened 2026-09-16 17:16 by Copilot

Add Arm KleidiAI compatibility documentation

#32642 opened 2026-09-16 14:02 by damdoo001-arm

u24 image override

#32635 opened 2026-09-16 07:39 by kpkbandi

Fix CUDA GQA graph replay test diagnostics

#32632 opened 2026-09-16 05:59 by Copilot

[MLAS] Fix fp16 overflow/NaN bugs in ARM64 NEON Gelu/Erf and add extreme-value test coverage

#32631 opened 2026-09-16 05:45 by kjg0724