Question Answering on SQuAD (test) (GPT-Judge Metrics) benchmark leaderboard · SOTA2 Research

Question Answering on SQuAD (test) (GPT-Judge Metrics)

89GPT Judge AccuracyCoT with SSRAG

Evaluation Results

Method	Links
CoT with SSRAG2026.01		89	79	80	0.2
CoVe with SSRAG2026.01		89	81.2	86	0.198
SSRAG2026.01		84	82.3	86	0.15
CoVe with SSRAG2026.01		84	77	83	0.13
CoT with SSRAG2026.01		83	83.4	82	0.13
SSRAG2026.01		83	78.9	93	0.27
SSRAG2026.01		82	82.3	90	0.21
SSRAG2026.01		81	81.2	94	0.21
CoT with SSRAG2026.01		81	71	87	0.28
CoVe with SSRAG2026.01		78	72	86	0.28
CoVe with SSRAG2026.01		77	74	87	0.22
CoT with SSRAG2026.01		76	76.5	85	0.23
SSRAG2026.01		74	85.2	96	0.29
CoT with SSRAG2026.01		74	77	84	0.29
CoVe with SSRAG2026.01		69	78	87	0.3
Graph RAG2026.01		66	58	65	0.37
Graph RAG2026.01		65	56	57	0.43
CoVe with Graph2026.01		63	57.7	58	0.39
Graph RAG2026.01		63	57	59	0.39
CoT with Graph RAG2026.01		62	54.7	63	0.33
Graph RAG2026.01		62	54	63	0.32
CoVe2026.01		58	48	41	0.47
CoVe with Graph2026.01		58	56	60	0.38
CoT with Graph RAG2026.01		57	58.2	66	0.39
CoVe with Graph2026.01		57	56	59	0.37
CoT2026.01		56	47	45	0.43
CoT with Graph RAG2026.01		56	59.9	69	0.32
CoVe with Graph2026.01		56	54	60	0.31
CoT with Graph RAG2026.01		55	59.8	69	0.3
Baseline RAG2026.01		54	43.2	47	0.52
Baseline RAG2026.01		53	49.9	51	0.53
Baseline RAG2026.01		53	49.7	56	0.41
CoVe2026.01		52	52.3	50	0.52
Graph RAG2026.01		52	49	58	0.41
CoT2026.01		51	47.8	51	0.49
Baseline RAG2026.01		51	52	54	0.42
CoVe2026.01		51	53	50	0.511
CoVe with Graph2026.01		50	57	58	0.39
CoT2026.01		50	54.3	50	0.43
CoVe2026.01		50	47.7	51	0.54
Baseline RAG2026.01		49	31.5	47	0.47
CoT with Graph RAG2026.01		49	59.9	66	0.41
CoT2026.01		49	49.8	51	0.46
CoVe2026.01		44	51	54	0.59
CoT2026.01		43	49.1	54	0.52