Information Retrieval on MAIR 38 Tasks benchmark leaderboard · SOTA2 Research

Information Retrieval on MAIR 38 Tasks

47Accuracy@1 (Average)GritLM-7B

Evaluation Results

Method	Links
GritLM-7B2025.10		47	44.1	58.2	43.3	42.6	57.6	40
E5-mistral-7B2025.10		46.8	45.4	55.4	42.3	43.1	55.3	40
ZeroGR-3B2025.10		41.1	42.7	47.4	40	38.3	39.2	36.3
OpenAI-Embed2025.10		40.6	40.6	48.2	31	39.7	49.4	28.7
BGE-Large2025.10		39.4	39.4	46.2	36	37.2	45.1	29
E5-Large2025.10		38.2	38.6	51	25	35.6	46.6	25.7
E5-Base2025.10		37.2	36.2	48.6	28.5	35.3	44.9	26.7
BGE-Base2025.10		37	38.6	40.2	25.8	37.6	42.2	29
BM252025.10		36.1	34.3	39.2	34.5	42.4	40	17.3
GTR-T5-large2025.10		35.4	39.8	39.6	27.8	31.8	38.5	24
Contriever2025.10		33.6	39.8	33.4	26.8	30.8	37.3	17.7
GTR-T5-base2025.10		32.5	36	33.6	25.3	31.9	37.4	18.7