List your company
ResearchFormerly WizWand

Interactive Decision Making on ScienceWorld

69.97Success RateMulti2
-2.278816.478135.23553.9919May 13, 2026May 17, 2026May 22, 2026May 26, 2026May 31, 2026Jun 4, 2026Jun 9, 2026
Updated 4mo ago

Evaluation results

MethodLinks
Multi2Base Model=Mistral 7B, Training Type=Fine-tuning-based, Structure Type=Hierarchical2026.0669.97
Multi2Base Model=Llama-3.1 8B, Training Type=Fine-tuning-based, Structure Type=Hierarchical2026.0667.61
HIPIFType=RL2026.0664.8
STEP-HRLType=RL2026.0661.8
Multi2Base Model=Qwen-2.5 3B, Training Type=Fine-tuning-based, Structure Type=Hierarchical2026.0660.68
GliderBase Model=Llama-3.1 8B, Training Type=Fine-tuning-based, Structure Type=Hierarchical2026.0660.48
GliderBase Model=Mistral 7B, Training Type=Fine-tuning-based, Structure Type=Hierarchical2026.0658.33
GIGPOType=RL2026.0658.1
GLIDERType=RL2026.0656.9
Gemini2.5proType=PE2026.0654.7
GliderBase Model=Qwen-2.5 3B, Training Type=Fine-tuning-based, Structure Type=Hierarchical2026.0654.69
Minimax-M2.7Paradigm=MAP, Model Category=Reasoning Models2026.0554.2
GPT-4o-1120Paradigm=MAP, Model Category=Non-reasoning Models2026.0553.3
Minimax-M2.7Paradigm=ReAct, Model Category=Reasoning Models2026.0551.1
PPOType=RL2026.0650.3
Minimax-M2.7Paradigm=CoMAP, Model Category=Reasoning Models2026.0550.2
Doubao-Seed-2.0-ProParadigm=MAP, Model Category=Reasoning Models2026.0548.9
GPT4oType=PE2026.0647.1
Kimi-K2-ThinkingParadigm=MAP, Model Category=Reasoning Models2026.0546.8
Hiagent+GRPOType=RL2026.0646.1
GPT-4o-1120Paradigm=CoMAP, Model Category=Non-reasoning Models2026.0544.8
Doubao-Seed-2.0-ProParadigm=CoMAP, Model Category=Reasoning Models2026.0544.1
Kimi-K2-ThinkingParadigm=CoMAP, Model Category=Reasoning Models2026.0543.5
GRPOType=RL2026.0643.1
Gemini2.5pro (Hiagent)Type=PE2026.0642.8
HiperType=RL2026.0642.6
Doubao-Seed-2.0-ProParadigm=ReAct, Model Category=Reasoning Models2026.0542.5
RL-GCDType=RL2026.0642.3
GPT-4o-1120Paradigm=ReAct, Model Category=Non-reasoning Models2026.0541.3
Kimi-K2-ThinkingParadigm=ReAct, Model Category=Reasoning Models2026.0541.2
Deepseek-V4-ProParadigm=MAP, Model Category=Reasoning Models2026.0541.2
MAP-4BParadigm=MAP, Model Category=Fine-Tuning Models2026.0540.5
GRPOBase Model=Mistral 7B, Training Type=Fine-tuning-based, Structure Type=Single2026.0638.47
RLOOType=RL2026.0638.4
GPT-4o-0806Paradigm=MAP, Model Category=Non-reasoning Models2026.0538.2
MAP-4BParadigm=CoMAP, Model Category=Fine-Tuning Models2026.0537.2
Deepseek-V4-ProParadigm=CoMAP, Model Category=Reasoning Models2026.0536.8
MAP-4BParadigm=ReAct, Model Category=Fine-Tuning Models2026.0535
GPT-4o-0806Paradigm=CoMAP, Model Category=Non-reasoning Models2026.0532.7
Qwen3-8B-ThinkingParadigm=MAP, Model Category=Reasoning Models2026.0532.4
Deepseek-V4-ProParadigm=ReAct, Model Category=Reasoning Models2026.0532.3
GPT4o (Hiagent)Type=PE2026.0632
GPT-4o-0806Paradigm=ReAct, Model Category=Non-reasoning Models2026.0531.4
Qwen3-8B-ThinkingParadigm=CoMAP, Model Category=Reasoning Models2026.0528.7
Qwen3-32B-ThinkingParadigm=MAP, Model Category=Reasoning Models2026.0528
ADaPTBase Model=Llama-3.1 8B, Training Type=Prompt-based, Structure Type=Hierarchical2026.0626.2
Qwen3-8B-ThinkingParadigm=ReAct, Model Category=Reasoning Models2026.0525.8
GRPOBase Model=Llama-3.1 8B, Training Type=Fine-tuning-based, Structure Type=Single2026.0625.79
Qwen3-4B-InstructParadigm=MAP, Model Category=Non-reasoning Models2026.0524.3
ReflexionBase Model=Mistral 7B, Training Type=Prompt-based, Structure Type=Single2026.0623.68
ACT-4BParadigm=MAP, Model Category=Fine-Tuning Models2026.0523.6
ReActBase Model=Mistral 7B, Training Type=Prompt-based, Structure Type=Single2026.0623.25
ReActBase Model=Llama-3.1 8B, Training Type=Prompt-based, Structure Type=Single2026.0623.23
ReflexionBase Model=Llama-3.1 8B, Training Type=Prompt-based, Structure Type=Single2026.0623.2
ReflexionBase Model=Qwen-2.5 3B, Training Type=Prompt-based, Structure Type=Single2026.0621.87
ReActBase Model=Qwen-2.5 3B, Training Type=Prompt-based, Structure Type=Single2026.0620.82
ACT-4BParadigm=CoMAP, Model Category=Fine-Tuning Models2026.0520.1
ADaPTBase Model=Qwen-2.5 3B, Training Type=Prompt-based, Structure Type=Hierarchical2026.0620
Qwen3-32B-ThinkingParadigm=CoMAP, Model Category=Reasoning Models2026.0519.7
GRPOBase Model=Qwen-2.5 3B, Training Type=Fine-tuning-based, Structure Type=Single2026.0618.11
ACT-4BParadigm=ReAct, Model Category=Fine-Tuning Models2026.0516.9
ReflexionType=PE2026.0616.4
Qwen3-4B-InstructParadigm=CoMAP, Model Category=Non-reasoning Models2026.0516.3
Qwen3-32B-ThinkingParadigm=ReAct, Model Category=Reasoning Models2026.0516.1
Qwen3-4B-InstructParadigm=ReAct, Model Category=Non-reasoning Models2026.0513.9
GPT-4o-miniParadigm=MAP, Model Category=Non-reasoning Models2026.0512.2
Qwen3-4B-ThinkingParadigm=MAP, Model Category=Reasoning Models2026.0511.4
HiagentType=PE2026.0610.8
ADaPTBase Model=Mistral 7B, Training Type=Prompt-based, Structure Type=Hierarchical2026.0610.5
ReactType=PE2026.0610.2
Qwen3-30B-A3B-InstructParadigm=MAP, Model Category=Non-reasoning Models2026.059.8
Qwen3-4B-ThinkingParadigm=CoMAP, Model Category=Reasoning Models2026.059.2
Qwen2.5-3B-InstructType=PE2026.067.8
GPT-4o-miniParadigm=CoMAP, Model Category=Non-reasoning Models2026.056.2
GPT-4o-miniParadigm=ReAct, Model Category=Non-reasoning Models2026.051.6
Qwen3-30B-A3B-InstructParadigm=CoMAP, Model Category=Non-reasoning Models2026.051.5
Qwen3-4B-ThinkingParadigm=ReAct, Model Category=Reasoning Models2026.051.5
Qwen3-30B-A3B-InstructParadigm=ReAct, Model Category=Non-reasoning Models2026.050.5