Predictive evidence reasoning on PredSet
10.4MAE (Overall)LITMUS (RE)AGENT
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| LITMUS (RE)AGENTAbbreviation=LRA2026.04 | 10.4 | 14.2 | 20 | 9.5 | 7.9 | 5.1 | 6.2 | |
| LITMUS++Abbreviation=L++2026.04 | 12.7 | 21.1 | 28.8 | 8.7 | 9.6 | 6.9 | 6.4 | |
| Single AgentAbbreviation=SA2026.04 | 14.5 | 17.8 | 29.6 | 12.4 | 11.9 | 4.7 | 7.5 | |
| ThoughtAgentAbbreviation=TA2026.04 | 15.1 | 18.9 | 27.5 | 14.7 | 11.6 | 7 | 9.9 | |
| GPT-4.1Abbreviation=GPT2026.04 | 16.5 | 39.8 | 18.2 | 9.6 | 9.9 | 6.8 | 8.3 | |
| Magentic-OneAbbreviation=M12026.04 | 17.2 | 27 | 26.8 | 14.1 | 12 | 8.3 | 10 |