microsoft/onnxruntime

Pull Requests Commits

Bump ruff from 0.12.12 to 0.17.4

dependabot[bot] committed 1 hour ago

Verified c6ce97c5

docs: refine JSEP migration build and rollout plan (#34092)

edgchen1 committed 6 hours ago

Verified e4234e56

onnxruntime-web - remove redundant native WebGPU adapter initialization (#34024)

edgchen1 committed 6 hours ago

Verified a14e230b

Revert "Verify downloaded Node packages (#33500)" (#34106)

eserscor committed 9 hours ago

Verified 0b22d27a

Read device memory type when adapting API tensors in EP adapter (#34101)

bmehta001 committed 9 hours ago

Verified 6208040e

Update Eigen to 5.0.1 + SSE2 fix (broken Clip & Sign) (#32475)

cbourjau committed 10 hours ago

Verified 85dcb542

[WebGPU] Handle absent optional bias in convolution and Attention (#33719)

gyagp committed 12 hours ago

Verified 79d8fdad

[CI] Split Windows Dawn builds from GPU tests (#34084)

eserscor committed 19 hours ago

Verified 6cc3c095

Bump plugin-ep-webgpu/VERSION_NUMBER to 0.7.0. (#34085)

edgchen1 committed 21 hours ago

Verified 4e1d7665

Bump onnxruntime version to 1.33 and cuda plugin ep to 0.5.0 (#34096)

tianleiwu committed 23 hours ago

Verified 2cfe9073

FlashAttention: use subgroup size 32 on Intel Xe2 and Xe3 (#33808)

jchen10 committed 1 day ago

Verified 47716d36

[WebGPU] Fuse QLoRA QKV and MLP projections (#33901)

qjia7 committed 1 day ago

Verified 1f307f81

Add BF16 support to CPU and CUDA Pad kernels (#34062)

Copilot committed 1 day ago

Verified adf4e971

Reject zero divisors in CUDA Mod kernels (#32546)

Copilot committed 1 day ago

Verified 9dd1d33d

[CUDA] Build the aarch64 wheel with family targets and fix CUDA arch gating (#34089)

tianleiwu committed 1 day ago

Verified bbedf7b1

[CUDA] Build FlashAttention for SM80 but not SM86/SM89 by default (#34088)

tianleiwu committed 1 day ago

Verified 82d47d09

[CUDA] Add MoEPacked CPU/GPU hybrid offload and consolidate QMoE/GGUF offload optimizations (#33973)

tianleiwu committed 1 day ago

Verified e3b9e0df

Constrain portable shared asset paths (#33826)

apsonawane committed 1 day ago

Verified 06d4aa36

[CUDA] Zero the cuDNN SDPA graph cache key before hashing it (#34081)

KafKafrnZ committed 1 day ago

Verified dbbdc8a1

Run Python tooling tests in CI (#34039)

edgchen1 committed 1 day ago

Verified 6998d9e0

Improve the app-EP weightless contract (#32899)

javier-intel committed 1 day ago

Verified d7ab0d27

[S0] Add internal capability planning protocol types (#33841)

chilo-ms committed 1 day ago

Verified e396ab51

[CUDA] Support FP16 KV cache and FP16 query/key in DynamicSparseAttention<BFloat16> and FP32 rotary cache / FP16 output in RMSNormRotary (#33997)

tianleiwu committed 1 day ago

Verified c93c80ff

Split Windows WebGPU static-plugin build/test jobs and cache vcpkg binaries (#34038)

eserscor committed 1 day ago

Verified 116967d5

Use CPU-only preparation build for Linux minimal CI (#33948)

eserscor committed 1 day ago

Verified 58878773

Fix GCC 14 build failure in graph topological sort (#34068)

Copilot committed 1 day ago

Verified 53934832

Extend MatMulPacked for native Bolt scaling and BF16 precision (#33962)

rui-ren committed 1 day ago

Verified 9566bec8

Speed up session creation (#33939)

cbourjau committed 1 day ago

Verified 24d81580

Add opt-in post-SiLU clipping to MoE operators (#33954)

baijumeswani committed 1 day ago

Verified d65da43f

Fix GQA shape inference for shape-derived int32 sequence lengths (#33884)

Copilot committed 1 day ago

Verified 3bde2541

Older