Multi-Query Attention failed to export onnx model

wenet-e2e / wenet

Production First and Production Ready End-to-End Speech Recognition Toolkit

https://wenet-e2e.github.io/wenet/

Apache License 2.0

4.08k stars 1.07k forks source link

Multi-Query Attention failed to export onnx model #2517

Closed John-Yao closed 4 months ago

John-Yao commented 4 months ago

I use Multi-Query Attention(implemented by repeat_interleave) in conformer, and export to onnx model. The onnx model failed to run by onnx-runtime. The onnx model seem to be truncated

B35F5A406F4DF10B4AEE603F692F5B27

Mddct commented 4 months ago

@John-Yao https://github.com/wenet-e2e/wenet/pull/2519 Please help verify

John-Yao commented 4 months ago

The onnx model run success by onnx-runtime. Nice!