List your company
ResearchFormerly WizWand

Mathematical & Geometric Reasoning on Geo3k

56.6Accuracy@8NoisyRollout-32B
36.3241.58546.8552.115Oct 10, 2025
Updated 8mo ago

Evaluation results

MethodLinks
NoisyRollout-32BModel Scale=32B, Prompting Source=Official author-provided, Note=Trained using the training set of Geo3k2025.1056.6
Qwen2.5-VL-32B + VPPOModel Scale=32B, Algorithm=VPPO2025.1053.4
NoisyRollout-7BModel Scale=7B, Training Strategy=Pure RL, Prompting Source=Official author-provided, Note=Trained using the training set of Geo3k2025.1051.8
Qwen2.5-VL-32B + GRPOModel Scale=32B, Algorithm=GRPO2025.1051.4
Qwen2.5-VL-32B + DAPOModel Scale=32B, Algorithm=DAPO2025.1051.4
MM-Eureka-32BModel Scale=32B, Prompting Source=Official author-provided2025.1051
Qwen2.5-VL-32BModel Scale=32B2025.1047
Qwen2.5-VL-7B + VPPOModel Scale=7B, Algorithm=VPPO2025.1046.5
PAPO-D-7BModel Scale=7B, Training Strategy=Pure RL2025.1044.1
Qwen2.5-VL-7B + DAPOModel Scale=7B, Algorithm=DAPO2025.1041.5
R1-ShareVL-7BModel Scale=7B, Training Strategy=Pure RL, Prompting Source=Official author-provided2025.1041.2
VL-Rethinker-7BModel Scale=7B, Training Strategy=Pure RL, Prompting Source=Official author-provided2025.1040.7
MM-Eureka-7BModel Scale=7B, Training Strategy=Pure RL, Prompting Source=Official author-provided2025.1040.3
Qwen2.5-VL-7B + GRPOModel Scale=7B, Algorithm=GRPO2025.1040.2
ThinkLite-7BModel Scale=7B, Training Strategy=Pure RL, Prompting Source=Official author-provided2025.1037.6
Qwen2.5-VL-7BModel Scale=7B2025.1037.1