Language-based Games on StrBuild (test)
1,620AdjacencyDiscussion
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DiscussionResponse time (seconds)=10.3, Cost (tokens/agent/turn)=2362026.01 | 1,620 | 6.5 | |
| ParallelResponse time (seconds)=9.4, Cost (tokens/agent/turn)=2322026.01 | 1,570 | 5.9 | |
| MAGRPOResponse time (seconds)=9.4, Cost (tokens/agent/turn)=2262026.01 | 1,330 | 50.6 | |
| PipelineResponse time (seconds)=9.8, Cost (tokens/agent/turn)=2462026.01 | 1,290 | 18.7 | |
| CoLLM-DCResponse time (seconds)=9.3, Cost (tokens/agent/turn)=1822026.01 | 760 | 44.6 | |
| CoLLM-CCResponse time (seconds)=9.5, Cost (tokens/agent/turn)=2392026.01 | 730 | 68.5 | |
| Raw ModelResponse time (seconds)=10.6, Cost (tokens/agent/turn)=4272026.01 | 90 | 36.6 | |
| GRPOResponse time (seconds)=10.3, Cost (tokens/agent/turn)=4112026.01 | 40 | 46.1 | |
| ACResponse time (seconds)=10.3, Cost (tokens/agent/turn)=4132026.01 | 40 | 49.8 |