search

ceccocats / tkDNN

Deep neural network library and toolkit to do high performace inference on NVIDIA jetson platforms
GNU General Public License v2.0
718 stars 209 forks source link

XavierNX convert yolo4tiny model to TensorRT FP16 engine got ERROR! #220

Closed jcyhcs closed 2 years ago

jcyhcs commented 3 years ago

jetson xavier nx , jetpack 4.4.1

i download yolo4tiny from github mainpage, then set export TKDNN_MODE=FP16 then exec ./test_yolo4tiny GOT ERROR: OH,PLEASE HELP ME!

mive93 commented 3 years ago

Hi @jcyhcs,

that is not a real error, the network will still work. You are using a quantized model (FP16) therefore there is a small error wrt the full precision one, but it's totally normal. If you use the demo, you will notice that the model works very fine.