microsoft/onnxruntime

Pull Requests Commits

Enable WebGPU CI for WebGPU plugin EP release branches (#32090)

edgchen1 committed 2 days ago

d3b1092b

[WebGPU] Add robustness provider option (#31971)

qjia7 committed 2 days ago

e4a2beaa

[WebGPU] Add Gather int64 support and make kernel version numbers function params (#31714)

miaobin committed 2 days ago

0272ba65

[CUDA] Add GatedAdd contrib operator (#31835)

tianleiwu committed 5 days ago

Verified 9b174a42

Validate SkipLayerNorm prepacked lengths (#31676)

apsonawane committed 5 days ago

Verified a29da166

Validate whisper beginning_timestamp_token_id bounds (#31636)

apsonawane committed 5 days ago

Verified 831fa2fe

Handle RNN activation parameters safely (#31675)

apsonawane committed 5 days ago

Verified 3cee1860

Harden CUDA fp16 transpose index math against int overflow (#31644)

apsonawane committed 5 days ago

Verified 37d79e38

Validate QLinearConv bias size against output channels (#31684)

apsonawane committed 5 days ago

Verified 45bdd4c4

Fix TfIdfVectorizer weight indexing semantics (#31649)

apsonawane committed 5 days ago

Verified f5de4c39

Validate FeatureVectorizer batch sizes (#31671)

apsonawane committed 5 days ago

Verified d3abafc3

[WebGPU] Add MRotaryEmbedding support (#31976)

qjia7 committed 5 days ago

Verified 1441b36d

Fix LinearAttention on GPUs with limited shared memory (#31982)

Copilot committed 5 days ago

Verified 86d38813

[WebGPU plugin EP] Update Python package test to only pass one WebGPU EP device (#31699)

edgchen1 committed 5 days ago

Verified 80f5546d

Validate GatherBlockQuantized CUDA indices bounds (#31645)

apsonawane committed 5 days ago

Verified 119daedf

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

Copilot committed 6 days ago

Verified e415ef9a

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

tianleiwu committed 6 days ago

Verified 16b486a2

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

tianleiwu committed 6 days ago

Verified af61f05a

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

chilo-ms committed 6 days ago

Verified caf31b56

Avoid eager static initialization in ORT (#31964)

mustjab committed 6 days ago

Verified eed2b754

Fix path traversal in TensorRT EP RefitEngine (#29396)

chilo-ms committed 6 days ago

Verified a31807ba

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

xiaofeihan1 committed 6 days ago

Verified 49193dbb

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

Honry committed 6 days ago

Verified 4855aea8

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

chilo-ms committed 6 days ago

Verified 808ba88b

Weightless support for all initializers (#29607)

chilo-ms committed 6 days ago

Verified 53211cd3

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

tianleiwu committed 6 days ago

Verified 1d69bbd6

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

tianleiwu committed 6 days ago

Verified 66eae3e6

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

mirounga committed 6 days ago

Verified 1eb1eb96

Fix GatherBlockQuantized default zero point on CUDA (#31693)

Copilot committed 6 days ago

Verified ab7d4609

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

adrastogi committed 6 days ago

Verified 33d509c7

Older