microsoft/onnxruntime

Pull Requests Commits

Allow optional misspell lint to fail

Copilot committed 5 days ago

Verified 959fd8ae

Copilot committed 5 days ago

Verified d18097f1

Add FP8 GEMV KSplit32 scheduling for the 48-SM SM121 GPU (#32409)

baijumeswani committed 5 days ago

Verified e9157780

Limit nested model graph depth (#32344)

apsonawane committed 5 days ago

Verified 79fc8c60

Register BFloat16 LayerNorm/RMSNorm kernels on the CPU EP (#31974)

justinchuby committed 5 days ago

Verified a6545f94

[WebGPU] Add int8 kv cache block quantization (#32284)

sushraja-msft committed 5 days ago

Verified c7c64085

Bump reviewdog/action-shellcheck from 1.31.0 to 1.32.1 (#32206)

dependabot[bot] committed 5 days ago

Verified a750fda2

perftest: Use plugin EP device allocators for input/output buffers (#32244)

jnagi-intel committed 6 days ago

Verified 31498663

webgpu: fix OOB subgroup-matrix loads for trailing partial tiles (#32364)

jchen10 committed 6 days ago

Verified 2a2833af

Fix FuseInitializersTransformer consumer input lookup (#32426)

sylvesterkaczmarek committed 6 days ago

Verified bb331b7a

Fix CUDA plugin user stream test lifetime (#32449)

Copilot committed 6 days ago

Verified 9f913ae5

[js/web] Support float16 in Conv3DNaive WebGPU shader (#32357)

Novestars committed 6 days ago

Verified 6690cf0d

[WebGPU] Share subgroup matrix MatMul with pointwise Conv (#32304)

qjia7 committed 9 days ago

Verified eebea690

[CUDA] Register BF16 ReduceMean kernels (#32326)

tianleiwu committed 9 days ago

Verified 3af6be47

Remove Hugging Face download dependency from test_dynamic_quantization_subgraph (#32443)

Copilot committed 9 days ago

Verified 781a3258

NetworkIsolation for NPM. Remove some unused parameters from release pipelines. Set default RC to 1. (#32440)

eserscor committed 9 days ago

Verified 208b435b

[MLAS] SVE Linear Attention Kernel (#32356)

mirounga committed 9 days ago

Verified 1756ea0d

Tune FP4 GEMV scheduling for the 48-SM SM121 GPU (#32408)

baijumeswani committed 9 days ago

Verified bae2f939

[CUDA] Add an opt-in FP8 DeepGEMM MoE decode path for QMoE (#32122)

tianleiwu committed 9 days ago

Verified aeb1a16a

[CUDA] Add speculative decoding to paged XQA (#32340)

tianleiwu committed 10 days ago

Verified 491a5f5f

Add packed attention workspace estimates (#32321)

titaiwangms committed 10 days ago

Verified 51df5b52

[CUDA] Optimize TopK for wide last axes (#32404)

tianleiwu committed 10 days ago

Verified 31ca9061

Reset plugin EP stream chunks before release (#31983)

apsonawane committed 10 days ago

Verified d47fd882

Bound cross_qk layer and head indices in CUDA Whisper beam search (#31998)

apsonawane committed 10 days ago

Verified 90e2f53b

Avoid pinned buffers in CUDA Split and Concat fast paths (#32410)

baijumeswani committed 10 days ago

Verified 2af39c97

Make workspace input shapes optional-aware (#32312)

titaiwangms committed 10 days ago

Verified f98e809f

Avoid overflow in CUDA reduction scans (#32137)

apsonawane committed 10 days ago

Verified bfa00fa6

Add AVX2 LayerNorm/RMSNorm kernel for x86-64 (MLAS) (#31973)

justinchuby committed 10 days ago

Verified 6dd9ba94

Add GetPreallocatedOutput to KernelContext. (#32089)

ericcraw committed 10 days ago

Verified e1bb372a

[WebNN] Fix inverted fallback data type support check (#32293)

Honry committed 10 days ago

Verified 66555cc4

Older