Garry Ling

m1 aarch64

why using GPU is much slower than CPU in this diffmpm.py example?

1

In my mac m1, using CPU(arm) diffmpm can running in 14FPS, but using GPU(metal) running much slower, only less then 2 FPS, also in 3080(cuda) ,is there any problem this...

[Description] For simply , I create an network only has transpose  then using nnapi backend to run, cause execution error.  this error hint ,the transpose parameter maybe invalid,...

feat(*): add neuron adapter api as draft

8

add neuro backend as draft

May some bug on custom layer...

3

## 🐞Describing the bug - this bug is quite hard to represent... - TLDR: - I create an custom layer, also convert success in coremltools, but show log of warning...

coreml using torch export API support maxpool

10

## 🌱 Describe your Feature Request coremltools export API failed to export maxpool, because MaxPool not adapt for "_parse_positional_args" and "_parse_keyword_args" parse for argument as "_convolution" ## How can this...

[BUG] <CUDA>crash on implicit gemm kernel

2

### Is there an existing issue for this? - [x] I have searched the existing issues ### Current Behavior I write a minimal demo to reproduce this crash ``` import...