End-to-End Question Answering on DoRA IBM Granite retriever (test) benchmark leaderboard · SOTA2 Research

End-to-End Question Answering on DoRA IBM Granite retriever (test)

57.52Token F1 ScoreDoRA SFT

Evaluation Results

Method	Links
DoRA SFT2026.04		57.52	57.23	40.57	0.038	89.85	66.46	4.21	29.33
Mistral-7B-Instruct2026.04		46	44.52	23.3	-0.223	86.44	57.59	6.22	36.19
Llama3.1-8B-Instruct2026.04		33.73	34.14	9.53	-0.39	82.84	60.75	5.75	33.49
Llama3-8B-Instruct2026.04		33.59	33.89	9.64	-0.393	82.9	60.52	5.45	34.02
GPT-3.5-Turbo2026.04		32.22	32.36	9.27	-0.273	83.3	62.72	4.25	33.03
GPT-4o2026.04		30.23	30.3	7.61	-0.316	81.99	66.1	5.11	28.8
GPT-4o-mini2026.04		25.35	25.52	5.78	-0.361	79.94	62.03	6.38	31.6
Claude-3-Haiku2026.04		22.4	22.51	7.05	-0.517	79.75	60.91	6.2	32.89
Qwen2.5-7B-Instruct2026.04		21.94	22.28	5.08	-0.444	78.84	59.69	6.56	33.74