microsoft/onnxruntime

Pull Requests Commits

perf(moe): parallelize QMoE CPU decode across active experts

tlwu committed 1 hour ago

20285f31

Avoid nested intra-op parallelism in QMoE CPU expert loop to fix livelock

Copilot committed 7 hours ago

Verified 854f9ef7

Copilot committed 8 hours ago

Verified 815bc8e8

Bump tar from 7.5.15 to 7.5.16 in /js/node (#29063)

dependabot[bot] committed 9 hours ago

Verified 84be9344

Bump js-yaml from 4.1.1 to 4.2.0 in /js (#29062)

dependabot[bot] committed 9 hours ago

Verified 6ae00279

webgpu: use naive reduction (#28174)

xhcao committed 19 hours ago

Verified a534347f

Harden ORT FlatBuffer model loader against malformed buffers (#28186)

tianleiwu committed 20 hours ago

Verified 9cacffa6

Bump tmp from 0.2.6 to 0.2.7 in /onnxruntime/test/wasm (#29055)

dependabot[bot] committed 1 day ago

Verified 693e090e

Add 2-bit quantization support to WebGPU GatherBlockQuantized operator (#29054)

Shivani767 committed 1 day ago

Verified 5c3c2b70

Fix OOB pointer in TensorAt for sub-byte packed types (#28973)

neilmsft committed 1 day ago

Verified f490d892

Bump WebGPU plugin EP version number to 0.3.0 (#29056)

edgchen1 committed 1 day ago

Verified 981b4757

Fix WebGPU GatherBlockQuantized dispatch failure for empty indices (#29030)

tairenpiao committed 1 day ago

Verified f67f7482

[WebGPU] Fix nearest half-tie handling for round_prefer_ceil/floor (#28757)

haoxli committed 1 day ago

Verified eabc40d9

[WebGPU] Enable Cast to int64 by default (#28804)

fanchenkong1 committed 1 day ago

Verified f0da3c4e

[WebNN EP] Support 8-bit MatMulNBits (#29025)

Honry committed 1 day ago

Verified c5aa86e1

Bump esbuild from 0.25.12 to 0.28.1 in /js (#29044)

dependabot[bot] committed 1 day ago

Verified bc226539

[CUDA] QMoE GEMV fast path for batch-1 decode (#29038)

tianleiwu committed 1 day ago

Verified 56f6fee6

Update protobuf to mitigate CVE-2026-0994 (#28967)

sushraja-msft committed 1 day ago

Verified 7e5c63ff

[WebNN EP] Remove unnecessary cast around normalization ops (#29024)

Honry committed 2 days ago

Verified bf7e27a4

Lower CUDA plugin EP minimum ORT version to 1.24.4 with version-gated callbacks (#28824)

tianleiwu committed 3 days ago

Verified 33b389aa

Add conditional logic to improve functionality (#29015)

apsonawane committed 4 days ago

Verified ed7a29b5

fix: NodeJS pkging stage needs to use CFS (#29006)

sanaa-hamel-microsoft committed 4 days ago

Verified d6e4a2d2

Use user-supplied external initializer in place when already on the planned device (#29013)

tianleiwu committed 4 days ago

Verified be41f5c6

[MLAS] Update the NHWC sans transposes path to also support Depthwise convolutions (#28565)

orlmon01 committed 4 days ago

Verified bb1dfff7

Implement Name Based partitioning and update documents (#28903)

yuslepukhin committed 4 days ago

Verified 29616be9

Add `OrtErrorCode` documentation (#29018)

edgchen1 committed 4 days ago

Verified 9c5e3be2

Update optimizer opset version checks for latest ONNX opset 26 (#28966)

yuslepukhin committed 4 days ago

Verified 0b278bbf

Fill CUDA opset gap for Softplus and Softsign to opset 22 (#28982)

elwhyjay committed 4 days ago

Verified 7db78930

Bump shell-quote from 1.8.3 to 1.8.4 in /js/react_native/e2e (#29022)

dependabot[bot] committed 4 days ago

Verified 0172bcd0

[CUDA] Optimize QMoE SoftmaxTopK router for small-batch decode (#28980)

tianleiwu committed 4 days ago

Verified dbf95cfb

Older