Nicklas Börjesson

Because with regards to messages, most fields are in the root of the item, an thus doesn't even need any xpath references, but can be inferred from correctly capitalizing the...

I have just realized that there are a bit too many questions that needs answering before I do this, my process will be too inefficient this way, so I will...

PEP8: Meh. With that attitude, you'll never get that little "LinkedIn"-company or whatchamacallit out of the basement. :-) Copyright: Me neither. Hm. One should really know about these things. Well,...

> Hi @nicklasb , > > Can you please help share your application changes done OR any minimal example based on IDF that we can use to replicate the issue...

@WrinkLeeTR I have it working (sloowly) with a 2 MB model, note that the for you, the tensor arena will probably take more like 2,5 MB in SPIRAM. i needed...

Oh, I just saw this here, great stuff.

@vikramdattu Hi, testing with an ESP32-P4 dev board (320 Mhz) and realized that there probably are not many P4-specific optimizations enabled? Because I have an inference (YOLOV5) that takes ESP32-S3...

@vikramdattu So now it is down to 35 seconds, so some improvement, certainly, however no where near 2x-7x, so as of yet, it does less per cycle than the S3,...

> Hi @nicklasb you're right. This is just a start. The optimizations are at an early stage, and that too only conv function optimized right now. This will definitely be...

> BTW, can you share with me the link for the S3 vs P4 documentation you've referred to? I have missed this particular point. Would help me understand the premise...