These Theta weights as described in the paper are the covariate mean adjustment effects that are saved in the alpha file. So, in the software we refer to these as alpha weights.
In sections 4.10.2 and 4.10.3 of the following manual:
you will see these described. Covariates are dummy-coded, such that the final level of categorical is the reference level.
For example, if you code Gender as 1=Male and 2=Female, then Female will be the reference "0" alpha values. So, in the alpha file with a covariates run you will find two sets of alpha utilities. These are labeled in the first row. The intercept utilities would be the Female utilities (the utilities when the covariate is at its reference level, meaning female). Then, after the intercept alpha utilities follow a set of alpha utilities associated with the adjustment to the intercept when the respondents are male. Thus, for a particular utility value, if the intercept is equal to 1.0 and the male alpha value is 0.5, then the population utility for male respondents is equal to 1.0 + 0.5 = 1.5.