List your company
ResearchFormerly WizWand

Interactive Decision Making on WebShop (test)

97Success RateBPO
1.52826.31451.175.886Aug 5, 2025Sep 7, 2025Oct 10, 2025Nov 12, 2025Dec 15, 2025Jan 17, 2026Feb 19, 2026
Updated 5mo ago

Evaluation results

MethodLinks
BPOApproach=System-2, Evaluation Protocol=Fine-Tuned, Base Model=Llama-3.1-8B-Instruct2025.0897—
Qwen-2.5-7B-InstructApproach=System-1, Evaluation Protocol=Zero-shot, Model Category=Base Models2025.0891.5—
ETOApproach=System-2, Evaluation Protocol=Fine-Tuned, Base Model=Llama-3.1-8B-Instruct2025.0890—
SFTApproach=System-2, Evaluation Protocol=Fine-Tuned, Base Model=Llama-3.1-8B-Instruct2025.0888—
MPOApproach=System-2, Evaluation Protocol=Fine-Tuned, Base Model=Llama-3.1-8B-Instruct2025.0887.5—
GiGPO + PA-MoEType=RL Training, Backbone=Qwen2.5-7B-Instruct2026.0282.893.1
GiGPO + PA-MoEType=RL Training, Backbone=Qwen2.5-1.5B-Instruct2026.0282.391
GEPOType=RL Training, Backbone=Qwen2.5-7B-Instruct2026.0280.591
o3-miniApproach=System-2, Evaluation Protocol=Zero-shot, Model Category=Reasoning LLMs2025.0876.5—
GEPOType=RL Training, Backbone=Qwen2.5-1.5B-Instruct2026.0275.689.9
GiGPO w/o stdType=RL Training, Backbone=Qwen2.5-7B-Instruct2026.0275.286.2
GiGPOType=RL Training, Backbone=Qwen2.5-7B-Instruct2026.0272.884.4
Qwen-3-8B-ThinkingApproach=System-2, Evaluation Protocol=Zero-shot, Model Category=Reasoning LLMs2025.0871.5—
PPO + PA-MoEType=RL Training, Backbone=Qwen2.5-7B-Instruct2026.0270.382.8
PPOType=RL Training, Backbone=Qwen2.5-7B-Instruct2026.0268.781.4
GRPO + PA-MoEType=RL Training, Backbone=Qwen2.5-7B-Instruct2026.0268.581.5
RLOO + PA-MoEType=RL Training, Backbone=Qwen2.5-7B-Instruct2026.0267.882.1
GiGPO w/o stdType=RL Training, Backbone=Qwen2.5-1.5B-Instruct2026.0267.483.5
GRPOType=RL Training, Backbone=Qwen2.5-7B-Instruct2026.0266.179.3
RLOOType=RL Training, Backbone=Qwen2.5-7B-Instruct2026.0265.780.3
GiGPOType=RL Training, Backbone=Qwen2.5-1.5B-Instruct2026.026583.1
Llama-3.1-8B-InstructApproach=System-1, Evaluation Protocol=Zero-shot, Model Category=Base Models2025.0860—
RLOO + PA-MoEType=RL Training, Backbone=Qwen2.5-1.5B-Instruct2026.0258.673.9
Deepseek-R1Approach=System-2, Evaluation Protocol=Zero-shot, Model Category=Reasoning LLMs2025.0858.5—
GRPO + PA-MoEType=RL Training, Backbone=Qwen2.5-1.5B-Instruct2026.025773.9
GRPOType=RL Training, Backbone=Qwen2.5-1.5B-Instruct2026.0256.875.8
PPO + PA-MoEType=RL Training, Backbone=Qwen2.5-1.5B-Instruct2026.0252.373.9
RLOOType=RL Training, Backbone=Qwen2.5-1.5B-Instruct2026.0252.173.9
PPOType=RL Training, Backbone=Qwen2.5-1.5B-Instruct2026.0251.573.8
Gemini-2.5-ProType=Prompting, Backbone=Closed-Source Model2026.0235.942.5
ReflexionType=Prompting, Backbone=Qwen2.5-7B-Instruct2026.0228.858.1
GPT-4oType=Prompting, Backbone=Closed-Source Model2026.0223.731.8
ReflexionType=Prompting, Backbone=Qwen2.5-1.5B-Instruct2026.0221.955.8
ReActType=Prompting, Backbone=Qwen2.5-7B-Instruct2026.0219.546.2
ReActType=Prompting, Backbone=Qwen2.5-1.5B-Instruct2026.0211.340.1
Qwen2.5Type=Prompting, Backbone=Qwen2.5-7B-Instruct2026.027.826.4
Qwen2.5Type=Prompting, Backbone=Qwen2.5-1.5B-Instruct2026.025.223.1