Here are parametric and nonparametric analysis results from the 2CEE JPL experiment.  They also have some COCOMO a,b information included.

In the "all" results the learners that stratified by Flight or Ground weren't considered (because this split the test instances).  So to see their performance look at the flight or ground results.

Apparently, running LOCOMO from the Alternate Estimation tab saves information about its a,b pairs correctly, but running it from the Normal tab does not.  This batch experiment was done using the Normal tab so I don't have information about LOCOMO a,b pairs.
Running the experiment and all analysis takes a very long time so I will only rerun that if there is high demand.

If you're curious about a,b pairs and LOCOMO, the calibration tab is useful, or use the alternate tab and look at the log file.