MMedbench
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
MMedbench English subset (val)
60.33Accuracy
36
MMedBench (test)
83.07Accuracy (Chinese)
20
MMedBench (test)
92.55Token Accuracy
16
MMedBench (test)
0.149Token Perplexity (log)
16
MMedBench (test)
69.9F1 Score
15
MMedBench 1.0 (test)
84.47Chinese Accuracy
9