Language Generation on AdaptSim
67.5Win RateBaseCRS (GPT-4o-mini) VS BaseCRS (3.5-turbo)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| BaseCRS (GPT-4o-mini) VS BaseCRS (3.5-turbo)Simulator Backbone=Qwen3-8B2026.06 | 67.5 | 27.2 | 5.2 | |
| BaseCRS (GPT-4o-mini) VS BaseCRS (3.5-turbo)Simulator Backbone=GPT-4.1-mini2026.06 | 60.2 | 38.5 | 1.2 | |
| BaseCRS (GPT-4o-mini) VS BaseCRS (3.5-turbo)Simulator Backbone=GPT-4o-mini2026.06 | 59.8 | 35.6 | 4.7 | |
| AgentCRS (GPT-4o-mini) VS AgentCRS (GPT-3.5-turbo)Simulator Backbone=Qwen3-8B2026.06 | 41 | 42.8 | 16.2 | |
| AgentCRS (GPT-4o-mini) VS AgentCRS (GPT-3.5-turbo)Simulator Backbone=GPT-4.1-mini2026.06 | 39.2 | 49.8 | 11 | |
| AgentCRS (GPT-4o-mini) VS AgentCRS (GPT-3.5-turbo)Simulator Backbone=GPT-4o-mini2026.06 | 36.2 | 50.2 | 13.5 | |
| BaseCRS (GPT-4o-mini) VS AgentCRS (GPT-4o-mini)Simulator Backbone=Qwen3-8B2026.06 | 33.2 | 60 | 6.8 | |
| BaseCRS (GPT-4o-mini) VS AgentCRS (GPT-4o-mini)Simulator Backbone=GPT-4o-mini2026.06 | 33 | 60 | 7 | |
| BaseCRS (GPT-4o-mini) VS AgentCRS (GPT-4o-mini)Simulator Backbone=GPT-4.1-mini2026.06 | 20.8 | 74 | 5.2 |