Why the loss appear in negative sign.?? on MAPNET

[DRAhmadFaraz]

@samarth-robo, as you can see in this screenshot, The loss turn into negative sign.? what is it mean.? when I try to train MAPNET on my GPU Nvidia RTX2080ti with these parameters as below

n_epochs = 300
batch_size = 20
do_val = yes
seed = 7
shuffle = yes
num_workers = 5
snapshot = 50
val_freq = 50
max_grad_norm = 0

[samarth-robo]

Hi, that is normal. The loss becomes negative because of the formula we use for the loss function.

On Thu, Aug 1, 2019 at 05:00 Ahmad Faraz Khan ***@***.***> wrote: @samarth-robo <https://github.com/samarth-robo>, as you can see in this screenshot, The loss turn into negative sign.? what is it mean.? when I try to train MAPNET on my GPU Nvidia RTX2080ti with these parameters as below n_epochs = 300 batch_size = 20 do_val = yes seed = 7 shuffle = yes num_workers = 5 snapshot = 50 val_freq = 50 max_grad_norm = 0 [image: Screenshot from 2019-08-01 16-55-29] <https://user-images.githubusercontent.com/38114191/62291499-8530f280-b47d-11e9-924b-0bc157b8e9cc.png> — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#21?email_source=notifications&email_token=AAVXKRR42NYMVQXENPRX5MDQCLF6TA5CNFSM4IIP45NKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HCZXGXQ>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAVXKRXTEJALYKJ3DIFDJPTQCLF6TANCNFSM4IIP45NA> .

-- Samarth Brahmbhatt https://samarth-robo.github.io/

[DRAhmadFaraz]

@samarth-robo Thanx for replying, I have one more question.

How many layers are there in mapnet Neural Network.? Can we change the layers? if Yes, How.?
?

[samarth-robo]

I don't remember exactly. I think we used resnet-34. All the network architecture definitions are in `models.py`

On Mon, Aug 5, 2019 at 23:28 Ahmad Faraz Khan ***@***.***> wrote: @samarth-robo <https://github.com/samarth-robo> Thanx for replying, I have one more question. How many layers are there in mapnet Neural Network.? Can we change the layers? if Yes, How.? ? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#21?email_source=notifications&email_token=AAVXKRSKYCQJGPFTRS24Z4LQDEK2NA5CNFSM4IIP45NKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD3UAYDA#issuecomment-518523916>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAVXKRSKJRFZZMF57O3WLILQDEK2NANCNFSM4IIP45NA> .

-- Samarth Brahmbhatt https://samarth-robo.github.io/

[DRAhmadFaraz]

@samarth-robo Sir I didnt get the "Models.py" file in your repository.

I only got Models/Posenet.py file and there no network is defined.

Regards

[samarth-robo]

Sorry, I meant models/posenet.py. The PoseNet class takes a feature_extractor as input, which acts as the "backbone" CNN. If you look at Line 76 of scripts/train.py, feature_extractor is set to resnet-34.