Coding Collaboration on CoopHE
75.2Pass@1CoLLM-CC
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| CoLLM-CCapproach=Multi-Agent Reinforcement Learning2026.01 | 75.2 | 75.9 | 77.8 | 86.5 | |
| MAGRPOapproach=Multi-Agent Reinforcement Learning2026.01 | 74.3 | 76 | 77.5 | 85.3 | |
| Pipelineapproach=Prompt-based multi-agent2026.01 | 62.5 | 76.4 | 77.3 | 85.3 | |
| ACapproach=Fine-tuning2026.01 | 62.2 | 62.6 | 63 | 63.3 | |
| GRPOapproach=Fine-tuning2026.01 | 61.8 | 62 | 62.2 | 62.8 | |
| CoLLM-DCapproach=Multi-Agent Reinforcement Learning2026.01 | 59.1 | 60.5 | 60.5 | 62.8 | |
| Single-Agentapproach=Base model2026.01 | 56.3 | 58.7 | 61.9 | 62.5 | |
| Parallelapproach=Prompt-based multi-agent2026.01 | 50 | 51.1 | 68.8 | 68.8 | |
| Discussionapproach=Prompt-based multi-agent2026.01 | 25 | 31.5 | 34.3 | 74.7 |