Francisco Melo

Hey @RangeKing , very nice work I just have a question... When I do: ``` net = torch.load('yolov8m.pt') print(net) ``` Only 1 Detection module appears... but in your image there...

Good idea!

UPDATE: I just tried to load the bitstream only to RAM, and it works: ``` openFPGALoader -c ft232 build/gsd_orangecrab/gateware/gsd_orangecrab.bit Jtag frequency : requested 6.00MHz -> real 6.00MHz Open file: DONE...

Exactly, if you're familiar with orangecrab and litex, my .bit includes gateware + firmware ;) And yes! I'm sorry, that command was actually `openFPGALoader -c ft232 -b colorlight -f build/image.bit`....

Very weird... I tried several times and in one of them the flash was recognized: ``` $openFPGALoader -c ft232 -f build/gsd_orangecrab/gateware/aa.bit write to flash Jtag frequency : requested 6.00MHz ->...

If the bitstream is alone, the flash works. But if in the same .bit I have the bitstream (gateware) + firmware, it doesn't work Yes I literally use jumper wires......

What about using the model? How can you be sure it's not against their terms? @musabgultekin @sowa705

Thanks for the reply @tuzhijun However, I still have some questions: 1. According to your paper, in order to compute binary activations (alpha_b) you say that the expression is eq....

Update: I renamed the bitstream file and: ``` init.. IDCODE: 0x41111043 (LFE5U-25) ECP5 Status Register: 0x06400000 reset.. flash ID: 0xEF 0x40 0x18 file size: 266307 erase 64kB sector at 0x000000.....

Makes sense, thanks for the reply ;) And the red bliking, what does it mean?