Open-domain Question Answering on NQ single-hop (test)
47.7Avg@16Dr. MAS
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Dr. MASBackbone=Qwen2.5-7B, Agent Configuration=Multi-Agent & LLM Non-Sharing2026.02 | 47.7 | — | — | — | 59.5 | |
| Dr. MASBackbone=Qwen2.5-7B, Agent Configuration=Multi-Agent & LLM Sharing2026.02 | 47.4 | — | — | — | 60.7 | |
| GRPOBackbone=Qwen2.5-7B, Agent Configuration=Single-Agent2026.02 | 46.4 | — | — | — | 57.6 | |
| GRPOBackbone=Qwen2.5-7B, Agent Configuration=Multi-Agent & LLM Sharing2026.02 | 45.2 | — | — | — | 60 | |
| Dr. MASBackbone=Qwen2.5-3B, Agent Configuration=Multi-Agent & LLM Non-Sharing2026.02 | 44.6 | — | — | — | 58.1 | |
| Dr. MASBackbone=Qwen2.5-3B, Agent Configuration=Multi-Agent & LLM Sharing2026.02 | 43.8 | — | — | — | 58.5 | |
| GRPOBackbone=Qwen2.5-3B, Agent Configuration=Multi-Agent & LLM Non-Sharing2026.02 | 43.8 | — | — | — | 54.5 | |
| GRPOBackbone=Qwen2.5-3B, Agent Configuration=Multi-Agent & LLM Sharing2026.02 | 41 | — | — | — | 59 | |
| GRPOBackbone=Qwen2.5-3B, Agent Configuration=Single-Agent2026.02 | 40.6 | — | — | — | 54.7 | |
| GRPOBackbone=Qwen2.5-7B, Agent Configuration=Multi-Agent & LLM Non-Sharing2026.02 | 27.1 | — | — | — | 39 |