Search results

Add local_window_size attribute to Attention op for sliding window support

Status: Open.

#7914 In onnx/onnx;

· justinchuby opened on May 4, 2026

feat: bump protobuf floor to 6.x (off EOL) and raise C++ standard to 20 — target 1.23 (Q3/26 release)

Status: Open.

#7904 In onnx/onnx;

· andife opened on May 3, 2026

[Feature Request] Add GroupedMatMul op for efficient MoE support

Status: Open.

#7902 In onnx/onnx;

· justinchuby-bot opened on May 2, 2026

Resize nearest half_pixel+round_prefer_ceil 20→6: ORT returns 0.7368 at element 4

bug

Status: Open.

#7841 In onnx/onnx;

· beanduan22 opened on Apr 15, 2026

QLinearMatMul: Reference impl. wraps instead of saturates out-of-bound values

bug

Status: Open.

#7835 In onnx/onnx;

· christeefy opened on Apr 15, 2026

Attention op (opset 24): add optional pre-softcap additive bias input

topic: operatorIssues related to ONNX operatorsIssues related to ONNX operators

Status: Open.

#7829 In onnx/onnx;

· justinchuby opened on Apr 13, 2026

Are Huntr reports being reviewed?

questionQuestions about ONNXQuestions about ONNX

Status: Open.

#7827 In onnx/onnx;

· xeloxa opened on Apr 11, 2026

https://huntr.com/repos/onnx/onnx

Status: Open.

#7818 In onnx/onnx;

· andife opened on Apr 6, 2026

LRN: Reference impl. iterates over incorrect axis

bug

Status: Open.

#7805 In onnx/onnx;

· christeefy opened on Apr 3, 2026

ConvTranspose: checker should validate that in_channels is divisible by group

bug

Status: Open.

#7802 In onnx/onnx;

· AyoubMDL opened on Apr 1, 2026

[Question] DequantizeLinear op with int32 input and zero point

questionQuestions about ONNXQuestions about ONNX

Status: Open.

#7752 In onnx/onnx;

· edgchen1 opened on Mar 17, 2026

CUDAExecutionProvider inserts MemcpyToHost/MemcpyFromHost around Resize for opset >=19 but not for opset 17/18

questionQuestions about ONNXQuestions about ONNX

Status: Open.

#7747 In onnx/onnx;

· ArtemVerbov opened on Mar 17, 2026