List your company
ResearchFormerly WizWand

Instruction Following on ALFWorld (val seen)

88.57Success Rate (SR)COMAP
-2.055621.47224568.5278Apr 27, 2026May 5, 2026May 14, 2026May 22, 2026May 31, 2026Jun 8, 2026Jun 17, 2026
Updated 4mo ago

Evaluation results

MethodLinks
COMAPBackbone=Qwen3-8B, Category=Co-evolving, Training-based=true2026.0688.57—
Imagine-and-ActBackbone=GPT-5.4, Category=API-Called Models, Training-free=true2026.0686.9—
TCOD (B2F)Student Model=Qwen2.5-7B2026.0486.4311.06
Imagine-and-ActBackbone=DeepSeek-V4-Pro, Category=API-Called Models, Training-free=true2026.0686.1—
Qwen2.5-7B-RL (Teacher)Model Type=Teacher, Role=Oracle/Upper Bound2026.0485.7110.61
ITPRBackbone=Qwen3-8B, Category=World Modeling, Training-free=true2026.0685.71—
ReActBackbone=GPT-5.4, Category=API-Called Models, Training-free=true2026.0683.2—
ReActBackbone=DeepSeek-V4-Pro, Category=API-Called Models, Training-free=true2026.0682.4—
TCOD (F2B)Student Model=Qwen2.5-7B2026.0482.1413.22
IWMBackbone=Qwen3-8B, Category=World Modeling, Training-free=true2026.0682.14—
TCOD (F2B)Student Model=Qwen2.5-3B2026.0481.4311.76
WKMBackbone=Qwen3-8B, Category=World Modeling, Training-free=true2026.0679.29—
TCOD (B2F)Student Model=Qwen2.5-3B2026.0477.8612.57
Vanilla OPDStudent Model=Qwen2.5-7B2026.0475.3713.18
COMAPBackbone=Qwen3-4B, Category=Co-evolving, Training-based=true2026.0671.53—
Imagine-and-ActBackbone=DeepSeek-V4-Flash, Category=API-Called Models, Training-free=true2026.0671.2—
Imagine-and-Act (w/ SFT)Backbone=Qwen3-8B, Category=Base Agent Planning, Training-based=true2026.0668.14—
ReAct (w/ SFT)Backbone=Qwen3-8B, Category=Base Agent Planning, Training-based=true2026.0667.86—
ReActBackbone=DeepSeek-V4-Flash, Category=API-Called Models, Training-free=true2026.0667.8—
ITPRBackbone=Qwen3-4B, Category=World Modeling, Training-free=true2026.0667.15—
Vanilla OPDStudent Model=Qwen2.5-3B2026.0465.7214.73
Imagine-and-Act (w/ SFT)Backbone=Qwen3-4B, Category=Base Agent Planning, Training-based=true2026.0665.71—
IWMBackbone=Qwen3-4B, Category=World Modeling, Training-free=true2026.0664.29—
TCOD-F2BStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-8B2026.0664.2917.65
SAGE-OPDStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-8B2026.0664.2918.14
Entropy-Aware OPDStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-8B2026.0663.5717.28
OPD (rollout 2)Student Model=Qwen3-1.7B, Teacher Model=Qwen3-8B2026.066317.51
ReAct (w/ SFT)Backbone=Qwen3-4B, Category=Base Agent Planning, Training-based=true2026.0662.86—
SAGE-OPDStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-32B2026.0662.8218.75
WKMBackbone=Qwen3-4B, Category=World Modeling, Training-free=true2026.0662.14—
OPD (rollout 2)Student Model=Qwen3-1.7B, Teacher Model=Qwen3-32B2026.0662.1418.17
BaseTeacher Model=Qwen3-8B2026.0661.4317.95
OPDStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-8B2026.066017.8
Entropy-Aware OPDStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-32B2026.066018.8
OPDStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-32B2026.0658.5718.89
SAGE-OPDStudent Model=Qwen3-0.6B, Teacher Model=Qwen3-8B2026.0657.8618.46
BaseTeacher Model=Qwen3-32B2026.0657.1418.94
TCOD-F2BStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-32B2026.0657.1419.42
OPD (rollout 2)Student Model=Qwen3-0.6B, Teacher Model=Qwen3-8B2026.0656.4318.76
TCOD-F2BStudent Model=Qwen3-0.6B, Teacher Model=Qwen3-8B2026.0656.4318.51
OPDStudent Model=Qwen3-0.6B, Teacher Model=Qwen3-8B2026.065518.53
SFTStudent Model=Qwen2.5-7B2026.0454.2918.92
TCOD-F2BStudent Model=Qwen3-0.6B, Teacher Model=Qwen3-32B2026.0654.2920.18
SAGE-OPDStudent Model=Qwen3-0.6B, Teacher Model=Qwen3-32B2026.0654.2919.85
Entropy-Aware OPDStudent Model=Qwen3-0.6B, Teacher Model=Qwen3-8B2026.0654.2618.12
OPDStudent Model=Qwen3-0.6B, Teacher Model=Qwen3-32B2026.0653.5719.49
OPD (rollout 2)Student Model=Qwen3-0.6B, Teacher Model=Qwen3-32B2026.0652.1420.28
Entropy-Aware OPDStudent Model=Qwen3-0.6B, Teacher Model=Qwen3-32B2026.0648.5720.87
ITPIBackbone=Qwen3-8B, Category=Test-time Evolving, Training-free=true2026.0641.43—
Imagine-and-ActBackbone=Qwen3-8B, Category=Base Agent Planning, Training-free=true2026.0632.86—
SFTStudent Model=Qwen2.5-3B2026.0432.1422.85
ITPIBackbone=Qwen3-4B, Category=Test-time Evolving, Training-free=true2026.0631.43—
RAPBackbone=Qwen3-8B, Category=Test-time Evolving, Training-free=true2026.0628.57—
BaseStudent Model=Qwen3-1.7B2026.0625.7124.63
SFTStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-8B2026.062524.58
RAPBackbone=Qwen3-4B, Category=Test-time Evolving, Training-free=true2026.0624.29—
ReflexionBackbone=Qwen3-8B, Category=Test-time Evolving, Training-free=true2026.0624.29—
ReflexionBackbone=Qwen3-4B, Category=Test-time Evolving, Training-free=true2026.0621.43—
Imagine-and-ActBackbone=Qwen3-4B, Category=Base Agent Planning, Training-free=true2026.0620—
ReActBackbone=Qwen3-8B, Category=Base Agent Planning, Training-free=true2026.0619.29—
ReActBackbone=Qwen3-4B, Category=Base Agent Planning, Training-free=true2026.0614.29—
Zero-ShotStudent Model=Qwen2.5-7B2026.049.2928.34
Zero-ShotStudent Model=Qwen2.5-3B2026.047.8628.73
SFTStudent Model=Qwen3-0.6B, Teacher Model=Qwen3-8B2026.062.8629.32
BaseStudent Model=Qwen3-0.6B2026.061.4329.65