Commits microsoft/onnxruntime

docs(quantization): clarify CPU parameter recommendations (#28415)

Rishi-Dave committed 64 days ago

Verified b40bdd80

fix: support Float16/BFloat16/Float8 in TensorArray custom op (#28335)

Rishi-Dave committed 64 days ago

Verified e6b3eb0e

Fix incorrect output for QDQ Resize with non-nearest interpolation modes under ORT_ENABLE_ALL (#28454)

Copilot committed 64 days ago

Verified 03245616

Pre-check ConstantOfShape output size against input initializer before constant folding (#28751)

Copilot committed 64 days ago

Verified 55975681

Fix regression in RandomForestClassifier binary classification predictions (#28685)

Copilot committed 64 days ago

Verified db641d65

fix(quantization): skip MaxPool during FP8 static quantization (#28488)

Rishi-Dave committed 64 days ago

Verified e2789cdf

Check for duplicate and null attributes in Model Editor API CreateNode (#28758)

skottmckay committed 64 days ago

Verified 72138269

Register Flatten as Direct8Bit op in Python QDQ static quantizer (#28340)

Rishi-Dave committed 64 days ago

Verified 9984c702

Reject QDQ Gemm→QGemm fusion when alpha != 1 with bias (#28131)

elwhyjay committed 64 days ago

Verified 58527303

Address numerical accuracy issues. (#28743)

yuslepukhin committed 64 days ago

Verified 18e3e86d

chore: bump version to 1.28.0 (#28765)

sanaa-hamel-microsoft committed 64 days ago

Verified 3ddbdfba

[WebGPU] Default non-debug validation mode set to disabled (#28679)

hariharans29 committed 64 days ago

Verified e25750d8

[CPU] Add fix for fastmath corner case with failing tests via sbgemm_neon_kernel (#28394)

JonathanC-ARM committed 64 days ago

Verified e0d99e27

fix the case where the buffer for past_state == present_state (#28753)

Guenther Schmuelling committed 64 days ago

Verified e26ad80e

Skip SetupDi device discovery if Win32k system calls are disabled (#28535)

shiyi9801 committed 64 days ago

Verified a8baf5c4

fix(quantization): validate bias scale in QDQ Conv → QLinearConv fusion (#28229)

Rishi-Dave committed 64 days ago

Verified 5c34495a

Use abseil for readable POSIX stack traces in debug builds (#28405)

tianleiwu committed 64 days ago

Verified 8da5e910

feat(quantization): add opset-21 block_size attribute to QDQ (#28522)

Rishi-Dave committed 64 days ago

Verified 8fcb725d

fix: preserve Q/DQ around Resize when mode is linear or cubic (#28414)

Rishi-Dave committed 64 days ago

Verified 7abc315f

Fix PRelu returning NaN for infinite inputs on CPU EP (#28750)

Copilot committed 65 days ago

Verified 9f81e2df

[MLAS][KleidiAI] Apply fp16 split independent fixes (#28720)

Laan33 committed 65 days ago

Verified 1e2b6394

Zero-copy I/O for plugin EPs with HOST_ACCESSIBLE memory (#28037)

ericcraw committed 65 days ago

Verified a827f9c3

[CoreML EP] Add GatherND builder (#28598)

maxwbuckley committed 65 days ago

Verified 21e9fe39

Fix ARM CPUIDInfo bounds handling for unknown CPU vendors (#28344)

Copilot committed 65 days ago

Verified d1135499

Honor disable_synchronize_execution_providers for CUDA graph replay (#28686)

tianleiwu committed 65 days ago

Verified 312f5249

[MLAS] Fix MatMulNBits CPU LUT GEMM path to apply optional bias (#28742)

hariharans29 committed 65 days ago

Verified 1077c77e

[WebGPU] Support models with opset 24 ops and KV-shared decoder layers (Gemma 4) (#28501)

feich-ms committed 65 days ago

Verified e1f27d15

Add LpNormalization support for CUDA Execution Provider (#28724)

apsonawane committed 65 days ago

Verified 5ddf568f

Unify path validation in EPs, address security gaps (#28725)

yuslepukhin committed 65 days ago

Verified 2e677fea

Add boundary checks for input validation (#28747)

apsonawane committed 65 days ago

Verified a28d46d3