microsoft/onnxruntime

Pull Requests Commits

tianleiwu committed 2 days ago

d8b00b68

tianleiwu committed 3 days ago

125d7ca8

address review feedback: inline per-channel flag, add defensive enforce, simplify casts

tianleiwu committed 3 days ago

db6ea250

Merge remote-tracking branch 'origin/main' into copilot/fix-qlinearconv-per-channel-zero-points

tianleiwu committed 3 days ago

1f1a0946

fix: support N-D weights with unit leading dims in MatMulNBitsQuantizer (#28178)

Rishi-Dave committed 3 days ago

Verified 2979bab1

Fix out-of-bounds read in CropBase: validate scale attribute length (#28399)

Copilot committed 3 days ago

Verified 50d04b92

CUDA Plugin Packaging Pipeline Fix TSAUpload and Timeout (#28497)

tianleiwu committed 3 days ago

Verified dcf87757

fix: auto-upgrade model opset to 21 for int16/uint16 QDQ quantization (#28202)

Rishi-Dave committed 3 days ago

Verified 0ebd8cc9

Bump next from 16.1.5 to 16.1.7 in /js/web/test/e2e/exports/testcases/nextjs-default (#27705)

dependabot[bot] committed 3 days ago

Verified 4d9aa47a

Address string attribute issue (#28494)

yuslepukhin committed 4 days ago

Verified 4bc20949

Optimize FlashAttention for M4 Max (20x speedup) (#27780)

xenova committed 4 days ago

Verified 938b6075

Address latent bug in LabelEncoder (#28496)

yuslepukhin committed 4 days ago

Verified 5f729560

Fill opset gap for RandomNormal, RandomNormalLike, RandomUniform, RandomUniformLike CUDA ops with BFloat16 support (#27759)

Copilot committed 4 days ago

Verified 0139499a

Fixes for unicode path handling (#28390)

jnagi-intel committed 4 days ago

Verified 9b3fb24a

Add float zero point support for 2-bit LUT GEMM in MatMulNBits (#28354)

vraspar committed 4 days ago

Verified e8ae6ce9

Bound total output allocation size in Tile kernel (#28070)

GopalakrishnanN committed 4 days ago

Verified a363e895

Fix OOB reads in SoftmaxCrossEntropyLoss via label bounds validation (#28004)

vraspar committed 4 days ago

Verified ad11f7cf

Use weights_only=True for remaining torch.load() calls (#28421)

adrianlizarraga committed 4 days ago

Verified 4e377a13

fix: use W_zero_point_is_uniform for PerColumnZeroPoints and rename depthwise fallback test

tianleiwu committed 4 days ago

bbbbcbb9

Add LabelEncoder_4 fast path for session creation (#27526)

SamuelLess committed 5 days ago

Verified 73d54165

Add RVV (RISC-V Vector Extension) optimized convolution and pooling kernels for the NCHWc blocked format in MLAS (#28411)

velonica0 committed 5 days ago

Verified 0e721889

Address Resize kernel shortcomings (#28402)

yuslepukhin committed 5 days ago

Verified 469bc9f6

[webgpu] Optimize LinearAttention Op with subgroup (#28412)

daijh committed 5 days ago

Verified a8260cd6

Fill CUDA EP opset gap for Sin and Cos operators (7→22) (#27756)

Copilot committed 5 days ago

Verified 480aad09

Use symbol file to export symbols in onnxruntime library on AIX platform as well (#28016)

ayappanec committed 5 days ago

Verified 239ef9e3

[WebGPU] Correct MatMul bias input indexing in `MatMul::ComputeInternal` (#28475)

Copilot committed 5 days ago

Verified 72bce1df

fix: skip DQ->MatMulNBits fusion when weight/scale initializer is shared (#28326)

Rishi-Dave committed 5 days ago

Verified 183e7a94

[WebGPU] Fix SkipSimplifiedLayerNormalization bias (#28427)

xenova committed 5 days ago

Verified fafe5644

Add CPU QMoE 2-bit support and LUT GEMM fast path (#28185)

tianleiwu committed 5 days ago

Verified dac5a143

[ Cuda] ConvTranspose-22 (#27710)

tianleiwu committed 5 days ago

Verified a12b611e

Older