Why does dream coder use e-graphs vs fragment grammars?
brando90 opened this issue · comments
I noticed that the original preprint uses fragment grammars but that was changed to use e-graphs in the PLDI accepted paper.
I was curious, why did that change happen? Was it that e-graph "worked better" or was it that the PL audience would be more welcoming of a PL/e-graph method for learning abstractions?
cross: CatherineWong/dreamcoder#1