microsoft/onnxruntime

Pull Requests Commits

Add fused MRotaryEmbedding contrib op for Qwen mRoPE variants (#31728)

Copilot committed 1 day ago

Verified e415ef9a

[Build] Fix cuda plugin windows wheel name (#31959)

tianleiwu committed 1 day ago

Verified 16b486a2

[CUDA] Add state_window to LinearAttention and CausalConvWithState to support MTP (#31157)

tianleiwu committed 1 day ago

Verified af61f05a

Remove onnxruntime/python/tools/tensorrt (unused internal dashboard tooling) (#29395)

chilo-ms committed 1 day ago

Verified caf31b56

Avoid eager static initialization in ORT (#31964)

mustjab committed 1 day ago

Verified eed2b754

Fix path traversal in TensorRT EP RefitEngine (#29396)

chilo-ms committed 1 day ago

Verified a31807ba

[WebGPU] Deferred-dispatch to parallelize cold-start shader compilation (#29557)

xiaofeihan1 committed 1 day ago

Verified 49193dbb

[WebGPU EP] Support int64 for Min and Max (#31709)

Honry committed 1 day ago

Verified 4855aea8

Wire max-shape inference into workspace estimation (#31613)

chilo-ms committed 1 day ago

Verified 808ba88b

Weightless support for all initializers (#29607)

chilo-ms committed 1 day ago

Verified 53211cd3

[CUDA] Split matmul nbits gemv to multiple cu files (#31837)

tianleiwu committed 1 day ago

Verified 1d69bbd6

[CUDA] Fix CUDA 13 packaging build failures (#31748)

tianleiwu committed 1 day ago

Verified 66eae3e6

[ARM] MLAS: portable machine-code SVE elementwise kernels, FEXPA exp (#31145)

mirounga committed 1 day ago

Verified 1eb1eb96

Fix GatherBlockQuantized default zero point on CUDA (#31693)

Copilot committed 1 day ago

Verified ab7d4609

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

adrastogi committed 1 day ago

Verified 33d509c7

[Build] Split matmul_nbits cu files (#31834)

tianleiwu committed 2 days ago

Verified 9d52a149

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

tianleiwu committed 2 days ago

Verified e9c76ade

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

elwhyjay committed 2 days ago

Verified 3162778d

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

bmehta001 committed 2 days ago

Verified 44895032

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

tianleiwu committed 3 days ago

Verified 2e76898a

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

kjg0724 committed 3 days ago

Verified 4c1d32b3

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

adrastogi committed 4 days ago

Verified 4dfcacc9

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

miaobin committed 4 days ago

Verified 31137265

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

tianleiwu committed 4 days ago

Verified 7843226c

Add regression test ConstantFoldingCopiesAliasedTensorBuffer for PR 29789 (#29826)

tianleiwu committed 4 days ago

Verified 1cea0d2e

Update CUDA plugin package outputs (#31722)

tianleiwu committed 4 days ago

Verified 776e8a81

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

hariharans29 committed 4 days ago

Verified 62aced3b

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

tianleiwu committed 4 days ago

Verified 1039416f

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

apsonawane committed 4 days ago

Verified d9b4c3fd

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

apsonawane committed 4 days ago

Verified c7e35639

Older