I have been following the transfer learning tutorial and have trained a few models such as EfficientNet-B2, ResNet-18, etc with accuracy ranging from 82-88% on a plant disease classification dataset for my capstone project. However, when I tried to use the MobileNetV3_small model. I noticed that there was a constant decline in accuracy.
The accuracy constantly dips and never seems to recover.
Hello. First time opening an issue on GitHub so I apologize if this is lacking something. I am currently using:
Jetson Nano Dev Kit 4 Gb. Python 3.6.9 Pytorch (torch) 1.10.0 TorchVision 0.11 JetPack SDK 4.6.1-b110 Linux4Tegra (L4T) Ubuntu 18.04.5 Docker 20.10.21 r32.7.4 Docker Container
I have been following the transfer learning tutorial and have trained a few models such as EfficientNet-B2, ResNet-18, etc with accuracy ranging from 82-88% on a plant disease classification dataset for my capstone project. However, when I tried to use the MobileNetV3_small model. I noticed that there was a constant decline in accuracy. The accuracy constantly dips and never seems to recover.