Multi-task Language Understanding on BIG-bench benchmark leaderboard · SOTA2 Research

Multi-task Language Understanding on BIG-bench

49.1AnachronismsOPT

Updated 5mo ago

Evaluation Results

Method	Links
OPT2022.11		49.1	19.8	52.9	55.3	36.8	34.1	48.8	21.1	44.6	52.5	20.6	88.5	43.5	19.1	69.2	13.6	78.6	12.9	42.3	57.1	21.3	49.6	35.3	79.2	49.5	52.2	31.6	34.4	54.9	46.2	55.1	57.5	57.5	54.4	52.6	49.8	46	12.5	26.7	32.9	51	42	42.8	36.2	42.1	38.7	57.1	20.8	60	22.4	41.4	48.8	58.5	28.4	22.2	19.6	39.7	43.4	42.7
Galactica2022.11		48.7	23.5	51.3	51.1	42.1	52.3	46.3	16.8	43	54.6	17.7	85.8	53	22.1	52.7	15.3	74.3	35.5	49.7	73.8	27.7	69.9	25.9	89.5	59.6	41.3	39.4	43.8	55.1	43.6	58	49.1	58	49.7	36.8	51.8	48	9.4	19.8	36.9	64	58	65.5	28	42.7	55	42.9	33	66.7	13.2	48.6	51.8	53.7	21.2	40.8	20.1	56.4	48.7	45.3
OPT2022.11		47.4	12.7	40	53.7	42.1	20.5	46.3	15.5	39.5	38.4	26.5	87.8	43.4	32.4	67.7	10.2	51.4	6.5	32.6	45.2	27.7	44.3	22.4	66.2	50.5	50	32.7	18.8	50.9	38.5	60.9	51.1	56.1	30.6	6.4	49.3	28	9.4	30.2	29.5	45	39.5	39.3	36.6	39.8	33.9	40.8	19.4	63.3	21.1	42	50	56.1	31.4	15.3	20.1	37.2	39.6	32.8
Galactica2022.11		47	17	47.1	49.5	42.1	40.9	48.8	11.4	46.9	50.5	26.5	89.2	40.1	25	66.7	13.6	68.6	16.1	36.6	71.4	19.2	59.4	16.5	87.8	60.6	50	35.8	34.4	54.1	25.6	44.9	63.4	51.6	52.3	31.6	50.9	38	6.3	12.8	40.9	53	55.6	55.9	34	41.1	45.8	46.9	24.1	60	21.1	48.1	52	53.9	26.4	39.9	28	56.5	46.6	42.7
BLOOM2022.11		1.3	19.2	48.6	54.7	47.4	31.8	51.3	12.2	44.2	56.6	26.5	89.2	61.2	29.4	65.2	22	80	22.6	48.6	59.5	19.2	53.7	28.2	89.5	63.6	50	31.1	25	54.5	30.8	52.2	61.5	54.8	57	49.4	51.1	48	15.6	22.1	28.2	55	37	54.2	39.6	5.2	40.5	44.9	12.5	65	27.6	47	54.5	57.1	20.5	24.4	24.9	38	42.6	42.2