task,baseline_reward,post_reward,delta,relative_delta_percent easy_syntax_fix,0.111850,0.123100,0.011250,10.06 medium_logic_fix,0.129350,0.101850,-0.027500,-21.26 hard_multi_bug,0.100600,0.100600,0.000000,0.00 hard_finance_explosion,0.103750,0.100413,-0.003337,-3.22 overall,0.111388,0.106491,-0.004897,-4.40