I had some problems with evaluating a custom test dataset on my fine-tuned model. So, I changed the train.py a little bit and added --evaluate argument to the parser. With this code, you can evaluate the test dataset:
python train.py --model cct_7_3x1_32 /path/to/test_dataset --resume /path/to/finetuned_model --evaluate
Also, when the task is classification with less than 5 classes, the validataion() function didn't work. So I fixed it with 21ac290 and 6b417f5 commits.
I had some problems with evaluating a custom test dataset on my fine-tuned model. So, I changed the train.py a little bit and added --evaluate argument to the parser. With this code, you can evaluate the test dataset:
python train.py --model cct_7_3x1_32 /path/to/test_dataset --resume /path/to/finetuned_model --evaluate
Also, when the task is classification with less than 5 classes, the validataion() function didn't work. So I fixed it with 21ac290 and 6b417f5 commits.