microsoft/onnxruntime

Pull Requests Commits

[webgpu] Use DoTranspose directly in Conv Op (#26074)

wenqinI committed 76 days ago

Verified b3ba5803

[webgpu] Remove 'webgpu' suffix from filenames (#26287)

daijh committed 76 days ago

Verified cd3aa49d

[webgpu] Fix graph capture broken (#26383)

qjia7 committed 76 days ago

Verified 949c0de7

Bump parkerbxyz/suggest-changes from 2 to 3 (#25839)

dependabot[bot] committed 76 days ago

Verified c7dfc25d

Bump vite from 6.3.6 to 6.4.1 in /js/web/test/e2e/exports/testcases/vite-default (#26370)

dependabot[bot] committed 76 days ago

Verified e7ef489d

[webgpu] Expand Squeeze version to 24 (#26359)

qjia7 committed 77 days ago

Verified b6b90a08

[webgpu] Simplify copyKVCache (#26371)

qjia7 committed 77 days ago

Verified 14f540b7

fix npm publishing (#26305)

guschmue committed 77 days ago

Verified 5ef8ce21

[WebGPU] Fix the typo `splitted` with `split` (#26360)

Jiawei-Shao committed 77 days ago

Verified e30f604b

[QNN-EP] Enable Optrace from QNN into QNN EP (#25987)

quic-calvnguy committed 77 days ago

Verified 50f07e27

CuFFTPlanCache: add destructor and Clear() for memory release (#26098)

cocotdf committed 77 days ago

Verified 50317ffb

Fix onnxruntime build on s390x with more recent GCC versions (#26374)

AlekseiNikiforovIBM committed 77 days ago

Verified 882c301b

[WebGPU] allows user to specify high-performance or low-power preference (#26326)

prathikr committed 77 days ago

Verified 00838cc7

[Web] Magic import comment missing for Vite (/* @vite-ignore */) (#25919)

mcost45 committed 77 days ago

Verified f5654ed8

[Build] New docker image with GCC 14 and CUDA 12.8/13.0 (#26349)

tianleiwu committed 77 days ago

Verified b970c118

Bump actions/setup-node from 5 to 6 (#26357)

dependabot[bot] committed 77 days ago

Verified ae71a573

[web] support JSPI build (#26095)

fs-eire committed 78 days ago

Verified 6efb844d

Fix quantization in Whisper model export (#26353)

jiafatom committed 78 days ago

Verified ed7847c1

[webgpu] Don't use num_workgroups when use indirect dispatch (#26334)

qjia7 committed 78 days ago

Verified 8ab27d9f

[build] disable DAWN_FETCH_DEPENDENCIES when using custom dawn src path (#26346)

fs-eire committed 78 days ago

Verified b1fdf9d3

[LARCH64 CPU]Provide inference acceleration optimization for Loongson CPU with 4-bit quantized models (#26280)

movedancer committed 81 days ago

Verified 7cc28b0b

[QNN EP] Reshape Transpose Fusion for 6D tensors (#26338)

quic-ashwshan committed 81 days ago

Verified 7804b5c1

S390x simd implementation (#25757)

AlekseiNikiforovIBM committed 81 days ago

Verified 510dd14f

Refactor: Move fundamental data type headers to core/common (#26330)

Changming Sun committed 81 days ago

Verified e0569fd9

Quant tool: Use 2D scales and zp in default MatMulNBitsQuantizer (#26329)

jambayk committed 81 days ago

Verified 6dbd41b5

[QNN EP] Upgrade QNN to 2.39.0 (#26249)

qti-jkilpatrick committed 81 days ago

Verified 878863d0

Quant tool: Don't store original outputs in MinMax Calibrator (#26321)

jambayk committed 82 days ago

Verified bbd0a80d

Internal Dupe of #25255 - [MLAS] Optimize MlasConv using thread partition opt (#26103)

hariharans29 committed 82 days ago

Verified 992c5981

Add support to filter_ep_devices using ep or device metadata in perftest (#26255)

jnagi-intel committed 82 days ago

Verified 9e79b367

[webgpu] Support RotaryEmbedding in flash attention (#26297)

xiaofeihan1 committed 82 days ago

Verified 90d92f82

Newer Older