Hello, I am an intern and I am using BRTs to model functional diversity of reef fish. I am testing several values of lr and tc together so I can find optimal settings for y models. for some of my data, I get the folowwing error message : "restart model with a smaller learning rate or smaller step size..." When I reduce lr or n.trees (with very small values), I still get the same error message. This is related to the following line in the brt.functions script: (cv.loss.values[j] > cv.loss.values[j-1])) the data drames I am working on contain around 60 to 120 lines and I have 51 predictors. I am new with BRTs and I have no idea how to resolve this issue. Any help would be appreciated!! Thanks Stephanie *--------------------------- Stéphanie D'agata* ECOSYM - UMR 5119 cc 93 Bat 24 - Université Montpellier II - Montpellier COREUS - UR 227 - IRD Nouméa - Nouvelle-Calédonie ** [[alternative HTML version deleted]]