Multilingual Commonsense Reasoning on M-Hellaswag benchmark leaderboard · SOTA2 Research

Multilingual Commonsense Reasoning on M-Hellaswag

79.2Accuracy (zh)CAUSAL-NATIVE

Evaluation Results

Method	Links
CAUSAL-NATIVE2025.05		79.2	83.3	73	80	74.6	72.1	74.1	63.1	74.9
CAUSAL-MULTI2025.05		77.2	84.6	74.3	80.6	75.3	70	74.8	66.4	75.4
MULTI-RELATED2025.05		76.6	76.9	77.6	76.1	74.7	62.6	70.3	69.1	73
MOREINFO2025.05		61.7	66	70.4	67.1	58.9	55.1	51.6	44.1	59.4
CAUSAL-NATIVE2025.05		59.3	53.4	58	61.3	67.9	45.7	55.2	65.7	56.4
CAUSAL-MULTI2025.05		59.3	65.8	64.1	63.1	55.8	54.7	51.9	51.3	58.3
CAUSAL-MULTI2025.05		58.8	56.4	58.9	62.6	67	57.4	59.7	68.5	61.2
MULTI-RELATED2025.05		57.1	64.5	58.5	57.1	56.7	52.9	45.5	46.1	54.8
CONFLICT2025.05		56.7	64.1	59.4	58.8	59.7	61.1	59.7	56.1	59.5
REFLECT2025.05		56	46.8	49.8	49.8	49.8	62.8	59.2	59.2	54.2
CAUSAL-NATIVE2025.05		55.8	56.8	58.5	59.2	42.4	51.5	39.4	42.5	50.8
MOREINFO2025.05		52.4	51.7	44.5	47.6	40.3	38.5	34.3	38.6	43.5
MULTI-RELATED2025.05		51.6	51.7	51.1	45.5	58	55.7	49.4	59.2	52.5
ASK CALI2025.05		51.1	50.4	52	54.1	48.1	53.8	49.8	49.6	51.1
MOREINFO2025.05		48.9	45.7	53.3	51.9	50.6	55.2	42.9	45.2	48.9
CONFLICT2025.05		48.5	53.8	51.5	51.1	—	48.4	49.8	49.6	50.1
REFLECT2025.05		48.1	48.3	50.7	48.1	46.3	50.2	47.2	50.9	47.7
ASK CALI2025.05		45.5	50	59	55.8	53.7	57	63.1	67.5	56.5
CONFLICT2025.05		41.6	41.7	50.7	47.1	61	58.5	56.1	59.9	52.1
ASK CALI2025.05		37.7	19.2	33.6	26.5	35.1	52.4	39.5	48.7	36.6
REFLECT2025.05		27.9	25.6	36.8	38.1	38.3	51.3	41.3	52.6	39