microsoft/onnxruntime

Pull Requests Commits

[CUDA] Honor deterministic compute for MatMulNBits tactic selection (#32898)

kunal-vaishnavi committed 3 days ago

Verified ed8f79bc

Avoid quadratic allocation-planner free-buffer scans (#33407)

rui-ren committed 3 days ago

Verified c33b4eb5

[CUDA] Cosmos3 - Fix variance cancellation in fused BERT layer normalization (#33357)

rui-ren committed 3 days ago

Verified f7b31b70

Fix CUDA LayerNorm/RMSNorm silent wrong output for rank-reduced scale (#28971)

justinchuby committed 3 days ago

Verified a926b837

Fix Gemm transpose fusion with public input outputs (#33484)

xgbtxy committed 3 days ago

Verified db5f5eaf

[WebGPU] Add bit-exact convolution state replay for speculative decoding (#33493)

aciddelgado committed 3 days ago

Verified 8b17e671

[Framework/CUDA] Add generic kernel dispatch policy reader (#33042)

chilo-ms committed 3 days ago

Verified 314ff8d4

Fix correctness guards in training graph optimizers (#32982)

Yuhx141 committed 4 days ago

Verified c55286e5

Validate ORT format edge slots (#32200)

apsonawane committed 4 days ago

Verified b923407c

[CICD Stability] Start transition to scaleset based providers (#33492)

eserscor committed 4 days ago

Verified 973f680e

[WebGPU] Automate Dawn updates (#33396)

gyagp committed 4 days ago

Verified 6a5a08c1

[WebGPU] Package D3D12 Agility SDK for Python wheels (#32795)

mei1127 committed 4 days ago

Verified fb8bfaaa

Update transformers and FlatBuffers Go example grpc dependency pins (#33501)

mustjab committed 4 days ago

Verified 53e61f3e

[CUDA] Add fused decode routing, single-pass finalize, and PDL overlap for NVFP4 QMoE (#33366)

tianleiwu committed 4 days ago

Verified 942356fc

Validate PackedAttention token offsets (#33330)

apsonawane committed 4 days ago

Verified a243f8cb

cuda pagedattention seqlens shape fix (#31651)

apsonawane committed 4 days ago

Verified 786177ab

Add opset 28 SpaceToDepth support (#32362)

justinchuby committed 4 days ago

Verified 4c184ac4

Enable BF16 for opset 28 operators (#32378)

justinchuby committed 4 days ago

Verified a9bf4811

Fix CUDA CICD. Remove unnecessary image override. (#33489)

eserscor committed 4 days ago

Verified ef1d8118

Update stale OrtEnv::m_ references in ort_env.cc comments (#33477)

mustjab committed 4 days ago

Verified 3d139a10

Validate quantization parameters before QDQ Add, Mul and Concat fusion (#33397)

rookieLiu2018 committed 4 days ago

Verified 192f45f9

Fix dynamic Gemm quantization with annotated or shared weights (#33415)

galleonli committed 4 days ago

Verified 22add83e

[WebGPU] Add opt-in shader cache collision diagnostics (#33153)

gyagp committed 5 days ago

Verified a9770bf1

[WebGPU] Add ApplyWideTileMatMulNBits helper (#33285)

daijh committed 5 days ago

Verified a7a1172c

Avoid module-load dynamic initializer for OrtEnv::m_ (#33419)

mustjab committed 5 days ago

Verified 28225b85

[WebGPU] Fix weight load acceleration so that weight load acceleration off session followed by on works (#33417)

sushraja-msft committed 5 days ago

Verified 9375dc30

Allow onnxruntime_DISABLE_EXCEPTIONS without onnxruntime_MINIMAL_BUILD (#33120)

rvandermeulen committed 5 days ago

Verified 9a662b93

[CUDA] Add opt-in tail-safe INT4 MatMulNBits dispatch for N % 8 == 4, M <= 4 (#33299)

tianleiwu committed 5 days ago

Verified bb68f5de

Skip MatMul BatchNormalization fusion when the channel axis moves (#33392)

samridh10exe committed 5 days ago

Verified dc1ae7a1

[CUDA] Add decode GEMV kernel for K-packed NVFP4 QMoE (#33354)

tianleiwu committed 6 days ago

Verified 69eeb2bb

Older