List your company
ResearchFormerly WizWand

Embodied Task Planning on VirtualHome (Seen)

31.2Success RateGRPO w/ EVU
6.2412.7219.225.68Apr 19, 2026
Updated 6mo ago

Evaluation results

MethodLinks
GRPO w/ EVUMethod Category=Training, Backbone Model=Qwen2.5-3B-Instruct, Base Method=GRPO, EVU Mechanism=Included2026.0431.2—————
PPO w/ EVUMethod Category=Training, Backbone Model=Qwen2.5-3B-Instruct, Base Method=PPO, EVU Mechanism=Included2026.0428.8—————
SFT w/ EVUMethod Category=Training, Backbone Model=Qwen2.5-3B-Instruct, Base Method=SFT, EVU Mechanism=Included2026.0427.2—————
GRPOMethod Category=Training, Backbone Model=Qwen2.5-3B-Instruct, Base Method=GRPO, EVU Mechanism=None2026.0425.6—————
PPOMethod Category=Training, Backbone Model=Qwen2.5-3B-Instruct, Base Method=PPO, EVU Mechanism=None2026.0424—————
GRPO w/ EVUMethod Category=Training, Backbone Model=Qwen3-1.7B-Instruct, Base Method=GRPO, EVU Mechanism=Included2026.0420—————
SFTMethod Category=Training, Backbone Model=Qwen2.5-3B-Instruct, Base Method=SFT, EVU Mechanism=None2026.0420—————
PPO w/ EVUMethod Category=Training, Backbone Model=Qwen3-1.7B-Instruct, Base Method=PPO, EVU Mechanism=Included2026.0417.6—————
ReAct w/ EVUMethod Category=Prompting, Backbone Model=DeepSeek V3.2, Base Method=ReAct, EVU Mechanism=Included2026.0416—————
SFT w/ EVUMethod Category=Training, Backbone Model=Qwen3-1.7B-Instruct, Base Method=SFT, EVU Mechanism=Included2026.0416—————
GRPOMethod Category=Training, Backbone Model=Qwen3-1.7B-Instruct, Base Method=GRPO, EVU Mechanism=None2026.0415.7—————
Plan-and-Act w/ EVUMethod Category=Prompting, Backbone Model=DeepSeek V3.2, Base Method=Plan-and-Act, EVU Mechanism=Included2026.0415.2—————
ReActMethod Category=Prompting, Backbone Model=DeepSeek V3.2, Base Method=ReAct, EVU Mechanism=None2026.0413.6—————
NoThinking w/ EVUMethod Category=Prompting, Backbone Model=DeepSeek V3.2, Base Method=NoThinking, EVU Mechanism=Included2026.0412.8—————
Plan-and-ActMethod Category=Prompting, Backbone Model=DeepSeek V3.2, Base Method=Plan-and-Act, EVU Mechanism=None2026.0412.8—————
PPOMethod Category=Training, Backbone Model=Qwen3-1.7B-Instruct, Base Method=PPO, EVU Mechanism=None2026.0410.4—————
NoThinkingMethod Category=Prompting, Backbone Model=DeepSeek V3.2, Base Method=NoThinking, EVU Mechanism=None2026.048—————
SFTMethod Category=Training, Backbone Model=Qwen3-1.7B-Instruct, Base Method=SFT, EVU Mechanism=None2026.047.2—————
finetuned GPT2 policybackbone=GPT-2, training_trajectories=10000, mode=fine-tuned2023.05—8,1305,9004,1203,090230
FLAREBackbone=Llama-3.2-3B2026.01—54.6919.93———
GPT3.5 Policybackbone=GPT-3.5, mode=few-shot, prompt_candidates=2002023.05—8,3404,7007,4304,820540
GPT3.5-MCTSbackbone=GPT-3.5, search=MCTS, mode=few-shot, prompt_candidates=2002023.05—9,1407,1208,8107,2603,360
LLM-PlannerBackbone=Llama-3.2-3B, Adaptation=Few-shot2026.01—50.9817.85———
LLM+FTBackbone=Llama-3.2-1B, Training=Fine-tuned2026.01—61.3717.98———
SayCanPaySay Backbone=Llama-3.2-3B, Pay Backbone=Llama-3.2-1B2026.01—64.1415.7———
TMoWBackbone=Llama-3.2-1B2026.01—83.6111.07———
UCTsearch=UCT, commonsense knowledge=false2023.05—00000
ZSPBackbone=Llama-3.2-3B, Adaptation=Zero-shot2026.01—10.7827.81———