Train loss curve understanding

Hello Everyone,
I am very new to ML field. I have used PyTorch to train my model and test.

I have plotted the curve as shown in the fig.

Can someone help me to give basic interpretation to the curve?

Thank You So Much.

@ptrblck Please help me out . I will really appreciate your help

The model curve tell us two things
Accuracy - Correct predictions
Loss - We reduce a loss using some kind of optimization.

Training accuracy is the one telling about correct samples predicted from training set
Validation acc tells about the predictions on validations set.

Seeing your model I assume it to be trained well as it’s validation and training accuracies are comparable as well loss curves.

1 Like

@Ali_Haider Thank you so much for your explanation :smiley: Really appreciate it.

Still you can further train it as your model might has some room for improvement. Models representational capacity is limited by optimization so, It depends if your model have achieved some local minima or like good minima. To see it better is to further train but save your current model.