WARNING: root: NaN or Inf found in input tensor.

[debparth]

I'm having very small dataset of only 100 Images in Training and 10 Images in Val.

My YML is: `project_name: technical_names train_set: train val_set: val num_gpus: 1

mean: [ 0.485, 0.456, 0.406 ] std: [ 0.229, 0.224, 0.225 ]

anchors_scales: '[2 ** 0, 2 ** (1.0 / 3.0), 2 ** (2.0 / 3.0)]' anchors_ratios: '[(1.0, 1.0), (1.3, 0.8), (1.9, 0.5)]'

obj_list: [ '1', '2' ] ` When I'm trying to train it with d0 model. It is giving me following error and not even training.

[zuofengyuan1]

Hello, I have the same problem, probably because the data set is too small. If you do not solve it, try the best data enhancement.

[debparth]

Then @zuofengyuan-web will 300 Images for Training would be good!?

[zylo117]

@debparth almost impossible, most likely overfitting

[debparth]

@zylo117 any idea how to solve this problem without overfitting? and also without getting this WARNING!?

[Muhibullah1]

I'm having very small dataset of only 100 Images in Training and 10 Images in Val.

My YML is: `project_name: technical_names train_set: train val_set: val num_gpus: 1

mean: [ 0.485, 0.456, 0.406 ] std: [ 0.229, 0.224, 0.225 ]

anchors_scales: '[2 ** 0, 2 ** (1.0 / 3.0), 2 ** (2.0 / 3.0)]' anchors_ratios: '[(1.0, 1.0), (1.3, 0.8), (1.9, 0.5)]'

obj_list: [ '1', '2' ] ` When I'm trying to train it with d0 model. It is giving me following error and not even training.

Reducing the batch size works for small dataset