Export saliency map #92
Conversation
There was a problem hiding this comment.
Looks like eval_model(L270) below in Inference_task.py should be updated together
: result, repr_vector = eval_model(return_loss=False, output_logits=True, **data)
The code change is making a side effect while training with OTE templates. Please check the issue with the comment below.
ote train ./external/mmsegmentation/configs/custom-sematic-segmentation/ocr-lite-hrnet-18-mod2/template_experimental.yaml --train-ann-files data/segmentation/custom/annotations/training --train-data-roots data/segmentation/custom/images/training --val-ann-files data/segmentation/custom/annotations/training --val-data-roots data/segmentation/custom/images/training --save-model-to data/seg-test params --learning_parameters.num_iters 2
No description provided.