Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Contextual Machine Translation on anaphora benchmark

0.54BLEU

gpt-4o

0.19680.28590.3750.4641Oct 20, 2025
Updated 1mo ago

Evaluation Results

MethodLinks
2025.10
0.540.710.920.91
2025.10
0.530.710.930.92
2025.10
0.490.70.920.92
2025.10
0.490.70.920.91
2025.10
0.490.670.90.9
2025.10
0.490.680.910.9
2025.10
0.490.680.920.91
2025.10
0.470.670.910.9
2025.10
0.450.670.920.91
2025.10
0.440.660.910.91
2025.10
0.440.660.920.9
2025.10
0.430.640.910.88
2025.10
0.390.590.840.84
2025.10
0.370.60.90.87
2025.10
0.350.590.890.87
2025.10
0.340.580.890.86
2025.10
0.340.580.890.86
2025.10
0.330.530.790.79
2025.10
0.30.540.870.85
2025.10
0.270.510.860.82
2025.10
0.270.50.850.8
2025.10
0.270.510.870.82
2025.10
0.240.50.860.8
2025.10
0.230.470.830.76
2025.10
0.210.450.770.77