List your company
ResearchFormerly WizWand

Multimodal Reasoning on SLAKE

90.62AccuracyDS-RL + GTP
69.029674.634880.2485.8452May 12, 2026May 20, 2026May 28, 2026Jun 5, 2026Jun 13, 2026Jun 21, 2026Jun 30, 2026
Updated 4mo ago

Evaluation results

MethodLinks
DS-RL + GTPModel=Lingshu, Scale=32B2026.0690.62
GTPModel=Lingshu, Scale=32B2026.0690.38
DS-RLModel=Lingshu, Scale=32B2026.0690.14
VISTA-SFTBackbone=InternVL3-8B2026.0587.61
R3VBackbone=InternVL3-8B2026.0587.04
STaR+Backbone=InternVL3-8B2026.0586.76
SFT-OracleBackbone=InternVL3-8B2026.0586.1
STaRBackbone=InternVL3-8B2026.0585.92
Instruct ModelModel=Lingshu, Scale=32B2026.0685.82
DS-RL + GTPModel=Lingshu, Scale=7B2026.0684.62
SFT-SeedBackbone=InternVL3-8B2026.0584.51
VISTA-SFTBackbone=InternVL3-2B2026.0582.54
DS-RLModel=Lingshu, Scale=7B2026.0682.45
VISTA-SFTBackbone=Qwen3-VL-2B-Instruct2026.0581.97
SFT-OracleBackbone=InternVL3-2B2026.0581.97
R3VBackbone=Qwen3-VL-2B-Instruct2026.0581.69
GTPModel=Lingshu, Scale=7B2026.0681.49
STaR+Backbone=InternVL3-2B2026.0581.41
Instruct ModelModel=Lingshu, Scale=7B2026.0681.25
SFT-OracleBackbone=Qwen3-VL-2B-Instruct2026.0580.28
R3VBackbone=InternVL3-2B2026.0580
STaR+Backbone=Qwen3-VL-2B-Instruct2026.0578.87
DS-RL + GTPModel=HuatuoGPT-Vision, Scale=7B2026.0678.61
DS-RLModel=HuatuoGPT-Vision, Scale=7B2026.0675.48
STaRBackbone=Qwen3-VL-2B-Instruct2026.0575.21
GTPModel=HuatuoGPT-Vision, Scale=7B2026.0674.52
STaRBackbone=InternVL3-2B2026.0571.55
SFT-SeedBackbone=InternVL3-2B2026.0571.54
Instruct ModelModel=HuatuoGPT-Vision, Scale=7B2026.0670.91
SFT-SeedBackbone=Qwen3-VL-2B-Instruct2026.0569.86