microsoft/onnxruntime

Pull Requests Commits

Fix GatherBlockQuantized default zero point on CUDA (#31693)

Copilot committed 10 days ago

Verified ab7d4609

Make DML runtime fused graph kernel own its model path (#31664)

adrastogi committed 11 days ago

Verified 33d509c7

[Build] Split matmul_nbits cu files (#31834)

tianleiwu committed 11 days ago

Verified 9d52a149

[Build] Reduce nvcc_threads to 1 for linux cuda plugin ci (#31729)

tianleiwu committed 11 days ago

Verified e9c76ade

Add double support for Cos op on CPU EP (#28975)

elwhyjay committed 11 days ago

Verified 3162778d

Minimize telemetry context and modernize 1DS integration (#29872)

bmehta001 committed 11 days ago

Verified 44895032

[Build] Install complete Perl runtime in Linux packaging images (#31724)

tianleiwu committed 12 days ago

Verified 2e76898a

[MLAS] Add int8 extreme-value coverage for ARM64 SymmQgemm (#31606)

kjg0724 committed 13 days ago

Verified 4c1d32b3

Validate constant tensor byte size in DML OnnxTensorWrapper (#31665)

adrastogi committed 13 days ago

Verified 4dfcacc9

WebGPU: Support int32 and uint32 for CumSum (#31702)

miaobin committed 13 days ago

Verified 31137265

[CUDA] Skip FP4 QMoE fc1 activation expansion (#31479)

tianleiwu committed 13 days ago

Verified 7843226c

Add regression test ConstantFoldingCopiesAliasedTensorBuffer for PR 29789 (#29826)

tianleiwu committed 13 days ago

Verified 1cea0d2e

Update CUDA plugin package outputs (#31722)

tianleiwu committed 13 days ago

Verified 776e8a81

[WebGPU] Initial PagedAttention implementation (1/n) (#31611)

hariharans29 committed 13 days ago

Verified 62aced3b

[CUDA] Add LinearAttentionGate and GatedRMSNorm contrib operators (#31158)

tianleiwu committed 13 days ago

Verified 1039416f

Use stream-aware scratch buffer in CUDA DeformConv (#31642)

apsonawane committed 13 days ago

Verified d9b4c3fd

Fix CUDA BeamSearch fp16 score mapping and add regression test (#31640)

apsonawane committed 13 days ago

Verified c7e35639

webgpu: Fix TurboQuant quantized KV cache for batch>1 with per-batch seqlens (#29752)

qjia7 committed 14 days ago

Verified cf5e9eba

Add more paths for cuda plugin ep (#31688)

tianleiwu committed 14 days ago

Verified a86d86ad

onnxruntime-web JSEP deprecation announcement for developers (#31683)

edgchen1 committed 14 days ago

Verified 407638ad

Add onnxruntime/test/providers/webgpu to plugin-ep-webgpu/paths.txt (#31685)

edgchen1 committed 14 days ago

Verified 2d1cfa02

[WebGPU] Support odd-N subgroup matrix MatMul weights (#29893)

jchen10 committed 14 days ago

Verified 1fe713be

Update ort release skill for cuda plugin (#31669)

tianleiwu committed 15 days ago

Verified 7ad118ca

[CUDA] GroupQueryAttention: dequantize the K and V caches in one launch (#31480)

tianleiwu committed 15 days ago

Verified c003602a

[WebGPU] Accumulate MatMulNBits wide-tile in output_element_t (#29611)

daijh committed 15 days ago

Verified baa0d32b

[CUDA] Speed up the NVFP4 QMoE decode GEMV and enable it for MTP verify (#31159)

tianleiwu committed 15 days ago

Verified e56b1073

[CPU] Tighten cache_indirection shape contract in MultiHeadAttention (#31634)

nenad1002 committed 15 days ago

Verified fbaee34d

Guard CUDA LayerNorm/RMSNorm int32 offset range (#31650)

apsonawane committed 15 days ago

Verified 24f1aedc

[CUDA] Add build option for TensorRT fused attention cubins (#31632)

tianleiwu committed 15 days ago

Verified 25cc21c2

Add ort-release-notes skill to draft release notes (#31614)

edgchen1 committed 15 days ago

Verified 1648b645

Newer Older