Machine Translation on M2M Language Family Subsets Low BLEU < 25 (test) benchmark leaderboard · SOTA2 Research

Machine Translation on M2M Language Family Subsets Low BLEU < 25 (test)

57.6AUS (9)M↔D

Evaluation Results

Method	Links
M↔D2025.01		57.6	70.7	66.9	55.9	72	—	64.5
M→D2025.01		57.2	73.9	76.2	55.6	72.2	51.2	64.4
M↔D2025.01		56.4	70.8	65.3	57.4	75.6	53.9	63.6
M→D2025.01		55.8	73.4	74.9	52	72.9	50.9	63.3
M↔D2025.01		55.5	73.6	73.8	54.1	77	55.6	61.8
D→M2025.01		54.2	70.3	57.8	55.2	74.2	51.1	58.5
fthrl2025.01		53.6	72.5	67.3	47	64.2	47.3	58.7
D→M2025.01		53.2	69.9	59.2	52.8	65	—	60
randaug-shell2025.01		53	72.1	67.7	47.3	64.2	47.3	58.6
off-the-shelf2025.01		52.9	72.6	67.9	47.6	64.2	47.6	58.8
D→M2025.01		52.3	73.1	66.7	50.5	75	—	60.5
off-the-shelf2025.01		46.9	71.5	44.7	47.6	64.2	47.6	53.8
cloud-cont2025.01		7.2	-2.2	15.3	8.3	7.8	8.5	8.4
cloud-all2025.01		6.6	-2	17.3	9.8	11.4	9.8	7.8
shell2025.01		5.1	1.5	19.4	8	8	3.6	7.6
all2025.01		4.4	-2.6	11.5	7.6	10	7.4	2.4
cloud-func2025.01		4.4	1.2	19.8	6.5	12.8	10.2	5.4
cloud2025.01		3.4	1	16.5	4.4	8.7	3.3	6.2
cont2025.01		2	-3.1	2.8	5.2	0.8	2.8	1.9
func2025.01		1	0.6	9	2.9	10.8	6.1	-0.1
ft-hrl2025.01		0.7	-0.1	-1.3	-0.6	0	-0.3	-0.2
randaug-shell2025.01		0.2	-0.5	-0.3	-0.3	0	-0.3	-0.2