microsoft/onnxruntime

Pull Requests Commits

Guard zero bias scale in FP16 QDQ quantization and expand Gemm tests

tianleiwu committed 1 day ago

14ad4dbe

Merge remote-tracking branch 'origin/main' into copilot/fix-quantization-issue

tianleiwu committed 1 day ago

1511e671

[WebGPU] Handle empty-rank MatMul and DequantizeLinear correctly (#33171)

mustjab committed 1 day ago

Verified dc894603

Fix OpenVINO provider info namespace for Python builds (#33327)

mustjab committed 2 days ago

Verified f8f8644c

Skip NCHWc conversion for zero-channel Conv (#33346)

rookieLiu2018 committed 2 days ago

Verified a5e205be

[WebGPU] Support 16x16x16 subgroup matrix MatMul (#32937)

SharmaRithik committed 2 days ago

Verified 9c87bdf2

[WebGPU] Preserve pending JSEP reads before CPU uploads (#33361)

gyagp committed 2 days ago

Verified 2db2a3ab

Add sm_110 (Jetson Thor) to the Linux aarch64 CUDA 13 packaging architectures (#33301)

LinCanNerd committed 2 days ago

Verified 6cbefccb

WebGPU: Fix DequantizeLinear 1-D block_size handling (#32734)

qjia7 committed 2 days ago

Verified 7c1103e2

[WebGPU] Validate MatMul execution plans in planner (#33288)

qjia7 committed 2 days ago

Verified bbfa896e

[WebGPU] Fix Pow scalar operands in vectorized broadcasting (#33348)

gyagp committed 2 days ago

Verified 9c21c823

Remove thread-pool parallelism assertion in GQA CPU test (#33342)

n1harika committed 2 days ago

Verified 07daf405

Fix Windows float16 model-path conversion (#33193)

aswanth-07 committed 3 days ago

Verified c9f4bb56

[MLAS] Skip FP16 unit tests on CPUs without FP16 arithmetic (#33260)

yairbenavraham committed 3 days ago

Verified 208463ad

Validate inline sparse payloads before dense allocation (#32863)

danielsongmicrosoft committed 3 days ago

Verified 9036f076

[WebGPU] Support fixed subgroup size in im2col_matmul (#33249)

Jiawei-Shao committed 3 days ago

Verified 6a78e30e

[CUDA] Include SM100 in Linux aarch64 CUDA 13 packages (#33315)

DevChuriwala committed 3 days ago

Verified e2e7a0a0

[CUDA] Add persistent fpA_intB MatMulNBits tactic autotune cache (#29588)

tianleiwu committed 3 days ago

Verified 42518d89

[float16] Reuse Cast nodes for shared always-float inputs (#33090)

Copilot committed 3 days ago

Verified 994f3497

Fix WebGPU pipeline callback test shader keys (#33310)

edgchen1 committed 3 days ago

Verified 63ff13be

[MLAS + CORE] Improve YOLOX FP32 models performance on CPU (#32335)

swetha097 committed 3 days ago

Verified 115e4499

Fix CPU OneHot handling of fractional floating-point indices (#33273)

Mohit-25-tech committed 3 days ago

Verified 9ddcd77f

Update libprotobuf-mutator to handle protobuf api changes (#33210)

eserscor committed 4 days ago

Verified da8e8973

[MLAS] Add support for Transposed B SME/SME2 Half Gemm path (#33130)

patryk-kaiser-ARM committed 4 days ago

Verified e0940a36

Reintegrate ONNX 1.23.2 base plumbing (#33134)

gramalingam committed 4 days ago

Verified 988cc07a

webgpu: Fix Python factory selection for plugin EP hosts (#33241)

qjia7 committed 4 days ago

Verified 67c61ae7

Remove stale Intel MKL third-party notice (#33233)

xadupre committed 4 days ago

Verified 86d60730

[WebGPU] Fix shader cache collisions across operator variants (#33242)

gyagp committed 6 days ago

Verified 8fc9fd59

[WebGPU] Prepack ConvTranspose1D/2D initializer weights (#33231)

gyagp committed 6 days ago

Verified 35714475

[WebGPU] Widen MatMulNBits SubgroupMatrix-8x16x16 tiling to 128x256 (#32605)

daijh committed 6 days ago

Verified bb47315d

Older