microsoft/onnxruntime

Pull Requests Commits

Merge finalized EPContext API into OpenVINO callback adoption

Gopalakrishnan Nallasamy committed 3 days ago

f28a3c4a

Address remaining EPContext API review refinements

Gopalakrishnan Nallasamy committed 3 days ago

46f94aa9

Keep stable EPContext callbacks independent of payload-size policy

Gopalakrishnan Nallasamy committed 3 days ago

a64e55dd

Account for stable EPContext APIs in Android size budget

Gopalakrishnan Nallasamy committed 3 days ago

166e2d6c

Replace old use of `routing_record` with `routing_snapshot_` in onnxruntime/contrib_ops/cuda/moe/moe_quantization.cc. (#32932)

edgchen1 committed 3 days ago

1b91bf72

Merge main and address stable EPContext API review feedback

Gopalakrishnan Nallasamy committed 5 days ago

9c93946c

[WebGPU] Fix initializer upload on GPUs without Resizable BAR (#32816)

SharmaRithik committed 6 days ago

Verified 96f73115

Expose CoreML EP via OrtEpFactory (#31975)

maxim-davgalev committed 6 days ago

Verified 626a7c56

Update MoE offload implementation plan and implements a global state for MoE/QMoE (#32738)

xadupre committed 6 days ago

Verified c2650ccb

[WebGPU] Keep LayerNormalization arithmetic in FP32 (#32896)

gyagp committed 6 days ago

Verified e927db3f

Use WRL COM pointers for DXCore device discovery (#32815)

mustjab committed 6 days ago

Verified 78c4614f

Reduce coordinate work in CPU N-D Im2col (#32817)

Hosi121 committed 6 days ago

Verified 38aee156

[MLAS] Avoid unused dynamic QGEMM packed buffer allocation (#32879)

mirounga committed 6 days ago

Verified d194fe39

Local non-causal for paged attention (#32792)

aciddelgado committed 7 days ago

Verified e1805779

Add github action for arm64 windows build (#32791)

eserscor committed 7 days ago

Verified e758c960

Validate sparse initializer value cardinality (#32753)

apsonawane committed 7 days ago

Verified 2c8379bc

Fix the nlohmann_json tarball hash in the standalone model_package build (#32596)

rvandermeulen committed 7 days ago

Verified 84cf511a

Use onnxruntime::narrow instead of gsl::narrow in core framework/graph/optimizer (#32591)

rvandermeulen committed 7 days ago

Verified 6b211684

Use ORT_TRY/ORT_CATCH for the remaining raw try/catch in the core non-minimal build (#32597)

rvandermeulen committed 7 days ago

Verified 20ba75f2

Fix Clang portability in logging, GELU, and spin pause (#32853)

mustjab committed 7 days ago

Verified 35597c6f

[WebGPU] Support concurrent use with explicit Session streams (#32587)

xiaofeihan1 committed 7 days ago

Verified 2b758bc6

Add packed INT2 and mixed-width CUDA QMoE decode (#32761)

jiafatom committed 7 days ago

Verified a11b4e59

Fix file placement for common linear attention file (#32831)

kunal-vaishnavi committed 8 days ago

Verified 92594005

test fix implemented (#32798)

melkap01-Arm committed 8 days ago

Verified cfad0a05

[CUDA] MatMulNBits: optional M chunking for the fpA_intB path (#32810)

tianleiwu committed 9 days ago

Verified 3756d4dc

[CUDA] Give each plugin EP session its own device arena (#32807)

tianleiwu committed 9 days ago

Verified 4c868628

Fix WebGPU build error when LTO is enabled (#32813)

skottmckay committed 9 days ago

Verified d4bd8135

Fix Python API docs ONNX dependency (#32811)

damdoo001-arm committed 9 days ago

Verified 3482ff20

Validate dynamic pads input access (#32654)

apsonawane committed 10 days ago

Verified f1fae4c9

Fuse BatchNormalization into ConvTranspose (#29542)

the0cp committed 10 days ago

Verified b61f007b

Older