Error using Logistic regression with one hot encoding for categorical predictors

Hi, I am new to MATLAB. I am trying to use logistic regression mnrfit on my data which contains several numeric and categorical variables. I have used one hot encoding to represent the levels of my categorical predictors.But every time I try to run mnrfit the following error shows up-

matrix is close to singular or badly scaled. results may be inaccurate. rcond = 1.256585e-22

and then-

Error using linsolve

Matrix must be positive definite.

I am guessing its beacause of all the dummy variables I have used to represent the categorical predictors with one hot encoding. What should I do?