MAXS: Meta-Adaptive Exploration with LLM Agents
Open paper
Benchmark
Model
Rank
Results
multimodal-reasoning-on-mathvista
MAXS (MiMo-VL-7B-SFT)
#13
Accuracy: 85.5
mathematical-reasoning-on-olympiadbench
MAXS (MiMo-VL-7B-SFT)
#30
Accuracy: 48.47
Rank counts only results with a code link.