Veeranjaneyulu Toka

Another time, but this time it is with bbox loss. Most likely if i am right, it is happening only with rpn (either bbox or class). Here is the log...

!!! Detected Infinity or NaN in output 0 of eagerly-executing op "Log" (# of outputs: 1) !!! dtype: shape: (1, 200, 1) #of -Inf elements: 100 Input tensor: tf.Tensor(****) I...

Hi @alexander-pv , Thanks for looking into this issue. I will consider latest changes and trigger training again with complete dataset. However i have tried earlier as well with coco_minitrain...

Hi, @alexander-pv , I wish you a very happy new year. I have started training with complete dataset, however still it is failing with Nan loss File "coco_train.py", line 98,...

Hi, @alexander-pv , FYI, i have decreased learning rate from 0.001 to 0.0001 and it is at 7th epoch as of now. Am not sure if this helps to resolve...

@b-nils I have been using this github to train coco weights (though not succeeded yet), but i found that using tf.data.Dataset api trains faster than that of tf.keras.utils.Sequence. I could...

@b-nils Might increase, but i faced an issue when i try to implement data pipeline using tf.data.Dataset API myself prior to refer to this github and i solved that issue...

Hi, @alexander-pv , Just to understand a bit more, why did you define prefetch() as an option in config? why it can not be a default behaviour?

Is preserving aspect ratio of an image (i.e. using tf.image.resize_with_pad) improves the model accuracy further?

Anybody has tried using mish and csp block in this repo?