文獻

TVM MetaSchedule - 分析operator的效能瓶頸

to-do

• [x] Send notion url
• [x] Send paper
• [ ] Profile LLM operator

Use transformer.js’s converter script to download and convert to onnx

Error encountered:

https://github.com/onnx/onnx/issues/4934

https://chatgpt.com/share/67f2ab7b-5eec-8009-820b-729b270128d1

So i would like to try on a smaller model whisper-small and mobilenet-v2 (CNN)

Fork this

一步一步解读神经网络编译器TVM(一)——一个简单的例子_tvm解读-CSDN博客

Inferencing with torch works

But inferencing with TVM fail for this reason