Sorry if this is a stupid question with an obvious answer but I’ve been messing around the example on github to train imagenet. Instead, I used it on my own custom dataset which is structured similarly. It seemed to train OK (could be better), but I don’t know how make the script run in evaluate mode. I know you use the -e flag for evaluate, but how do I tell it to use my newly created PTH file? I guess I could try writing my own code, but since the github example already has an eval mode…
You could pass the checkpoint path as the --resume argument and set --evaulate to just run the validation method.
I assume you’ve already added your model definition etc. to the script.
Would that work for you or am I missing something?
Thanks, I hadn’t thought of that. I’m still a novice. I trained my model and got a PTH file, so now my question is “now what?” I guess it’s time to evaluate.
“I assume you’ve already added your model definition etc. to the script.”
I’m not sure what you mean. I use the -a flag to make sure it’s using the right architecture (resnet101, in my case).