microsoft/onnxruntime

Pull Requests Commits

eserscor committed 26 days ago

827e9091

Document Stable Diffusion Transformers constraint

eserscor committed 31 days ago

8ae7b9bb

Run transformer CUDA tests in build container

eserscor committed 31 days ago

9d54ae97

Temporarily run transformer model tests on PR

eserscor committed 31 days ago

285c63d9

Fix Python 3.12 Stable Diffusion dependencies

eserscor committed 31 days ago

f1c66be2

Update transformer model dependency validation

eserscor committed 31 days ago

e2f87dc9

eserscor committed 31 days ago

a139b207

Bump transformers to 5.5.0

eserscor committed 31 days ago

9bac6497

Isolate bundled Dawn implementation from WebGPU context (#32964)

mustjab committed 31 days ago

Verified d74a6b97

Bump moment from 2.30.1 to 2.31.0 in /js/react_native/e2e (#32956)

dependabot[bot] committed 31 days ago

Verified 02f16929

[CUDA] Remove memset nodes from decode CUDA graphs and prefer GEMV at small M (#32884)

tianleiwu committed 31 days ago

Verified 06f675b7

Preserve EP registration End events for returned errors without telemetry (#32780)

bmehta001 committed 32 days ago

Verified b2097437

[CUDA] Fix packed-INT QMoE expert-counting capture and consume (#32942)

aciddelgado committed 32 days ago

Verified 74810e56

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

edgchen1 committed 32 days ago

Verified 80291e14

[CUDA] Replace shared ones-buffer bias broadcasts in Gemm and DecoderAttention (#32852)

Copilot committed 32 days ago

Verified 8ba5bec8

[CUDA] Auto-tune small-M decode MatMul (cuBLAS vs small-N GEMV) and speed up MatMulNBits GEMV (#32876)

tianleiwu committed 32 days ago

Verified 567deb38

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

SharmaRithik committed 32 days ago

Verified 96f73115

Expose CoreML EP via OrtEpFactory (#31975)

maxim-davgalev committed 32 days ago

Verified 626a7c56

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

xadupre committed 32 days ago

Verified c2650ccb

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

gyagp committed 32 days ago

Verified e927db3f

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

mustjab committed 33 days ago

Verified 78c4614f

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

Hosi121 committed 33 days ago

Verified 38aee156

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

mirounga committed 33 days ago

Verified d194fe39

Local non-causal for paged attention (#32792)

aciddelgado committed 33 days ago

Verified e1805779

Add github action for arm64 windows build (#32791)

eserscor committed 33 days ago

Verified e758c960

Validate sparse initializer value cardinality (#32753)

apsonawane committed 33 days ago

Verified 2c8379bc

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

rvandermeulen committed 33 days ago

Verified 84cf511a

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

rvandermeulen committed 33 days ago

Verified 6b211684

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

rvandermeulen committed 33 days ago

Verified 20ba75f2

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

mustjab committed 33 days ago

Verified 35597c6f

Older