CGA-bench/analysis/paper_runs/paper_fsm/paired_deltas.csv

1	model	repeat	task_id	baseline_coverage	cga_coverage	coverage_delta	baseline_semantic_coverage	cga_semantic_coverage	semantic_coverage_delta	baseline_eval2	cga_eval2	eval2_delta	baseline_time_sec	cga_time_sec	baseline_token_cost	cga_token_cost
2	qwen-max	1	2012_q2fsm	91.66666666666666	91.66666666666666	0.0	61.17	61.17	0.0	0.7	0.3	-0.39999999999999997	348.97	489.72	0.62682	0.7213599999999999
3	qwen-max	1	2013_q2afsm	92.3076923076923	92.3076923076923	0.0	73.51	73.51	0.0				519.09	1803.02	0.7850400000000001	2.66142