10 Cross validation and training and testing set?

Question

[Chaimaa]

Hi experts, I have used this function to extract the significant genes and used 10 cross-validations to find the best model, however when i run this command & it produces 59 models (different lambda ) and one MSE(minimum square error). Then I get confused about which lambda represent the best model or do i have to split the data into training and testing set?

I run again this command 10 times and i get different MSE.

i can't interpret this results well, Anyone already has an idea about this plz help me! this is the function I used from MATLAB/ [B,fitinfo] = lasso(x,y,'CV',10,'Alpha',0.5);