Share your thoughts, 1 month free Claude Pro on usSee more

Clarifying Questions on SciQA (test)

26Accuracy

Swift

Updated 1mo ago

Evaluation Results

Method	Links
Swift 2025.06		26	62	4.63	5.77	5.39	4.44	5.07	4.96
Refit 2025.06		24	68	5.03	6.21	5.81	4.75	5.42	5.33
STaR-GATE-D 2025.06		14	16	4.01	4.95	4.76	3.67	4.3	4.65
DPO 2025.06		8	24	4.44	5.53	5.17	4.19	4.86	4.86
StepDPO 2025.06		8	34	4.7	5.85	5.36	4.44	5.09	5.13
Base 2025.06		6	86	5.47	6.69	6.14	5.1	5.95	5.87