MiniCPM-SALA: Hybridizing Sparse and Linear Attention for Efficient Long-Context Modeling
Open paper
Benchmark
Model
Rank
Results
mathematical-reasoning-on-aime24
MiniCPM-SALA (9B)
–
Acc: 83.75
multi-task-language-understanding-on-mmlu-pro
MiniCPM-SALA (9B)
–
Accuracy: 67.04
Rank counts only results with a code link.