Hello & how do I report a memory leak in the TPU?

[seahope]

Hello everyone.

I was inspired by Jdaie Lin (PiSugar) to build his chatbot using the M5StackLM-8850AI M.2 Acceleration Card (AX8850). I found it works just as his video said :-)

I extended his chatbot project and enabled switching between a text and vision model. I knew that would take a while, but what I have found is that normally it works ok for 2 perhaps 3 model switches and then it fails with what seems to be to memory errors.

I was just wondering where the best place to log this issue is for the devs to fix?

Regards

S