microsoft/onnxruntime

Pull Requests Commits

address comments from Copilot

qjia7 committed 1 day ago

419e2630

Add component support to SplitPackedQKV

qjia7 committed 1 day ago

ceea4d88

make SplitPackedQKV work on GQA

qjia7 committed 1 day ago

efdc783d

SplitPackedQKV with BSD format

qjia7 committed 1 day ago

0067a2d9

remove debugging codes

qjia7 committed 1 day ago

cbb8ef22

qjia7 committed 2 days ago

dfb98e3f

change the splitQKV to BNSH format

qjia7 committed 4 days ago

8d38f51b

[webgpu] Optimize attentionPrepare

qjia7 committed 5 days ago

e15bd722

Verify in-memory references when loading a model (#26764)

yuslepukhin committed 13 days ago

Verified a83a1583

[webgpu] Add `CeilDiv` into webgpu utils (#26723)

daijh committed 14 days ago

Verified 334bc041

[VitisAI] Add support compiled model compatibility information retrieval and validation (#26699)

mingyueliuh committed 14 days ago

Verified b6d914b2

[TRTRTX EP] Support custom-ops used in NVFP4 recipe (#26555)

vishalpandya1990 committed 14 days ago

Verified 0318be22

[webgpu] Propagate rotary_interleaved parameter to GQA shader (#26758)

xenova committed 14 days ago

Verified d1abad00

Write parameters based on output shapes not the amount of names (#26092)

TedThemistokleous committed 14 days ago

Verified 283f1d3a

[QNN-EP] - Tie HTP power config id lifetime to session (#26457)

quic-calvnguy committed 14 days ago

Verified 8b81d9b1

[webgpu] Optimize Attention by enhancing flash attention support (#26715)

qjia7 committed 15 days ago

Verified 549d7415

Fix missing include for CudaHasBF16Support in conv_fp16_test.cc (#26671)

Copilot committed 15 days ago

Verified 07bf9a04

[TensorRT RTX EP] Be able to specify aux streams (#26569)

hadiFute committed 15 days ago

Verified 0b8180e2

[TRT RTX EP] Don't register set device function when we use existing stream (#26542)

hadiFute committed 15 days ago

Verified 655b69ca

[Migrphax] Fix docker file (#26737)

tianleiwu committed 15 days ago

Verified 935affb8

Bump next from 15.4.7 to 15.4.8 in /js/web/test/e2e/exports/testcases/nextjs-default (#26719)

dependabot[bot] committed 18 days ago

Verified 790018d2

[ROCM] remove more rocm ep code (#26712)

tianleiwu committed 18 days ago

Verified c34d9b07

Adjust Quantization Preprocessing on Opset Handling (#26503)

qti-hungjuiw committed 18 days ago

Verified 3e8be0e8

[QNN-EP] Remove outdated comment for per-channel quantization (#26523)

chunghow-qti committed 18 days ago

Verified 948bba1a

[EP ABI] Initial support for kernel-based EPs (#26206)

adrianlizarraga committed 18 days ago

Verified db6d83b6

[OVEP] Update OV version to 2025.3.0 (#26556)

preetha-intel committed 18 days ago

Verified 8a433fb2

[VitisAI] Add External EP Loader (#26627)

BoarQing committed 18 days ago

Verified 10afbefd

Make sure TRT EPs can loads models when initializers in memory (#26721)

yuslepukhin committed 18 days ago

Verified 66c9f1c0

[webgpu] Get data with `GetByOffset()` in `gemm_utils.cc` (#26679)

Jiawei-Shao committed 19 days ago

Verified 838cf03c

[webgpu] Fix the check of `ValidateVariableDataType()` with Split-K (#26725)

Jiawei-Shao committed 19 days ago

Verified 3c60bd52

Older