and welcome to the community!
Loss refers to the „ error“ or „deviation“ between the prediction of a model and the ground truth (label).
When fitting your ML model (training) you want to minimize your loss so that the model can predict well the labels (e.g. predict that there is a cat on a picture when there is really a cat visible).
Feel free to check out this thread: Difference between loss function and cost function? - #2 by Christian_Simonis
Please let me know if anything is unclear.
Best
Christian