List your company
ResearchFormerly WizWand

Success Rate (%) on ALFWorld Seen

85.7Success Rate (%)GRPO w/ EVU
35.15648.27861.474.522Apr 19, 2026
Updated 6mo ago

Evaluation results

MethodLinks
GRPO w/ EVUMethod Category=Training, Backbone Model=Qwen2.5-3B-Instruct, Base Method=GRPO, EVU Mechanism=Included2026.0485.7
GRPOMethod Category=Training, Backbone Model=Qwen2.5-3B-Instruct, Base Method=GRPO, EVU Mechanism=None2026.0483.6
PPO w/ EVUMethod Category=Training, Backbone Model=Qwen2.5-3B-Instruct, Base Method=PPO, EVU Mechanism=Included2026.0479.3
PPOMethod Category=Training, Backbone Model=Qwen2.5-3B-Instruct, Base Method=PPO, EVU Mechanism=None2026.0477.8
SFT w/ EVUMethod Category=Training, Backbone Model=Qwen2.5-3B-Instruct, Base Method=SFT, EVU Mechanism=Included2026.0470
SFTMethod Category=Training, Backbone Model=Qwen2.5-3B-Instruct, Base Method=SFT, EVU Mechanism=None2026.0465.7
ReAct w/ EVUMethod Category=Prompting, Backbone Model=DeepSeek V3.2, Base Method=ReAct, EVU Mechanism=Included2026.0456.4
ReActMethod Category=Prompting, Backbone Model=DeepSeek V3.2, Base Method=ReAct, EVU Mechanism=None2026.0455.7
NoThinking w/ EVUMethod Category=Prompting, Backbone Model=DeepSeek V3.2, Base Method=NoThinking, EVU Mechanism=Included2026.0455
Plan-and-Act w/ EVUMethod Category=Prompting, Backbone Model=DeepSeek V3.2, Base Method=Plan-and-Act, EVU Mechanism=Included2026.0453.6
Plan-and-ActMethod Category=Prompting, Backbone Model=DeepSeek V3.2, Base Method=Plan-and-Act, EVU Mechanism=None2026.0452.1
GRPO w/ EVUMethod Category=Training, Backbone Model=Qwen3-1.7B-Instruct, Base Method=GRPO, EVU Mechanism=Included2026.0452.1
NoThinkingMethod Category=Prompting, Backbone Model=DeepSeek V3.2, Base Method=NoThinking, EVU Mechanism=None2026.0450.7
PPO w/ EVUMethod Category=Training, Backbone Model=Qwen3-1.7B-Instruct, Base Method=PPO, EVU Mechanism=Included2026.0447.1
GRPOMethod Category=Training, Backbone Model=Qwen3-1.7B-Instruct, Base Method=GRPO, EVU Mechanism=None2026.0447
PPOMethod Category=Training, Backbone Model=Qwen3-1.7B-Instruct, Base Method=PPO, EVU Mechanism=None2026.0442.1
SFT w/ EVUMethod Category=Training, Backbone Model=Qwen3-1.7B-Instruct, Base Method=SFT, EVU Mechanism=Included2026.0441.4
SFTMethod Category=Training, Backbone Model=Qwen3-1.7B-Instruct, Base Method=SFT, EVU Mechanism=None2026.0437.1