We have simulated the Telescope Array FADC fluorescence detecter and surface detector array dataset from Jan. 2008 through Nov. 2018 with CORSIKA generated MC sets using three high-energy interaction models: QGSJetII-04, EPOS-LHC, and Sibyll 2.3d, and with four separate nuclear species for each model: hydrogen, helium, nitrogen and iron. The simulation includes detector response characteristics both of the surface detector, done using GEANT4, and for the fluorescence detectors, using ray-tracing, atmospheric attenuation and trigger simulation. Using the model and species data in energy bins, we created template $X_{\rm max}$ distributions, and used these templates to find the fractions of each species that best match the hybrid data for each model. The fractional fit was performed using a Markov Chain MC method which allows sampling of the likelihood space of the fractions. We present the fractions for each model and show the correlations between the fit fractions.