microsoft/onnxruntime

Pull Requests Commits

Fix CUDA tactic-cache environment lookups across provider boundaries

Copilot committed 24 days ago

Verified d5acf24f

Merge PR #29588 tactic-cache source for focused follow-up

Copilot committed 24 days ago

Verified 715dbe3a

Copilot committed 24 days ago

Verified 450834a7

Add reference examples and guidance (especially for multi-device scenarios) for ValidateCompiledModelCompatibilityInfo (#29168)

adrastogi committed 24 days ago

Verified dba3629f

Add WebGPU INT2 and mixed-width QMoE support (#32878)

sayanshaw24 committed 24 days ago

Verified ec1d573c

Fix Python WHL builds for older CUDA version. Filter out __grid_constant__ from tinygemm2 (#33111)

eserscor committed 24 days ago

Verified fc8f71cf

Promote EPContext data callbacks to stable API (#32265)

GopalakrishnanN committed 24 days ago

Verified 772123e9

Address review feedback on fpA_intB tactic cache

tianleiwu committed 24 days ago

91d0f38c

Harden fpA_intB tactic cache file handling

tianleiwu committed 24 days ago

9aac11e8

Fix tactic cache keying, plugin flush, and rebase regressions

tianleiwu committed 24 days ago

35b3a012

Address review findings in fpA_intB tactic cache

tianleiwu committed 24 days ago

b60f7690

[CUDA] Add persistent fpA_intB MatMulNBits tactic autotune cache

tianleiwu committed 24 days ago

5e2a1b55

[CUDA][WebGPU] Add packed sparse attention indexer for continuous batching (#32618)

Copilot committed 24 days ago

Verified 3188a2ee

Validate RestorePadding token offsets (#33099)

apsonawane committed 24 days ago

Verified 0ea42b2b

Raise ir_version to 10 when MatMulNBitsQuantizer emits INT4/UINT4 tensors (#33068)

andife committed 24 days ago

Verified f152edba

Fix LayerNorm fusion with zero normalized dimensions (#33079)

hulkbig committed 24 days ago

Verified 43ed141a

Validate PackedMultiHeadAttention token offsets (#33098)

apsonawane committed 24 days ago

Verified ca3e9482

[MLAS] Build the SVE asm generator's input TUs as C++20 (#33047)

vijay-kapse committed 24 days ago

Verified e8a4eb54

Add sliding window AVX-512 NCHWc depthwise kernel (bitwise identical) (#32749)

mirounga committed 24 days ago

Verified 9f4b6f4f

Fix regression in free threaded python wheel builds introduced with onnx rollback (#33105)

eserscor committed 24 days ago

Verified 8eef62d5

Upgrade transformers to 5.3.0 (#32752)

danielsongmicrosoft committed 24 days ago

Verified 0fddc3fc

Add onnxruntime_DISABLE_DEVICE_DISCOVERY to skip platform device probing (#32592)

rvandermeulen committed 25 days ago

Verified b734c811

WebGPU static plugin EP (#32395)

edgchen1 committed 25 days ago

Verified a84958ac

Bump github/codeql-action/autobuild from 3.37.7 to 4.38.2 (#33024)

dependabot[bot] committed 25 days ago

Verified d9b8ddc0

Filter model package output options (#33096)

apsonawane committed 25 days ago

Verified b0609a04

Validate GatherBlockQuantized output dimensions (#33097)

apsonawane committed 25 days ago

Verified 003599e1

Validate DLPack capsules before conversion (#33100)

apsonawane committed 25 days ago

Verified 99f13d09

Bound antialias resize coefficient windows (#33095)

apsonawane committed 25 days ago

Verified 27800d7f

Limit AOT function expansion (#32662)

apsonawane committed 25 days ago

Verified 2f3c9186

[CUDA] Validate GEMM autotuning candidates using graph replay (#33085)

tianleiwu committed 25 days ago

Verified ba9d0320

Older