Hi, thanks for the great paper and for open-sourcing the code!
Section 4.2 mentions that RuleFlow generated 120 rewrite rules, of which 88 passed qualitative correctness review and were used for the PandasBench evaluation. I don't see these rules included in the repository, and was wondering if you could share the 88 evaluated rules?
This would be really useful for reproducing your evaluation without re-running snippetgen and rulegen with PandasBench.
Even a JSON dump of the LHS/RHS/precondition triples would be very helpful. Thanks!
Hi, thanks for the great paper and for open-sourcing the code!
Section 4.2 mentions that RuleFlow generated 120 rewrite rules, of which 88 passed qualitative correctness review and were used for the PandasBench evaluation. I don't see these rules included in the repository, and was wondering if you could share the 88 evaluated rules?
This would be really useful for reproducing your evaluation without re-running
snippetgenandrulegenwith PandasBench.Even a JSON dump of the LHS/RHS/precondition triples would be very helpful. Thanks!