Cross-Domain Evaluation on Aggregate Generalization
40.2Overall Average ScoreECHO
Evaluation Results
| Method | Links | |
|---|---|---|
| ECHOBackbone=Qwen3-32B-Instruct2026.06 | 40.2 | |
| ECHO w/ Top-K retrievalBackbone=Qwen3-32B-Instruct2026.06 | 38.2 | |
| ECHO w/ Top-K retrieval & w/o turn summaryBackbone=Qwen3-32B-Instruct2026.06 | 36.8 | |
| ECHO w/o Traceable CA & Turn-level ISBackbone=Qwen3-32B-Instruct2026.06 | 36.1 | |
| ECHO w/o traceable CABackbone=Qwen3-32B-Instruct2026.06 | 35.6 | |
| SUPOBackbone=Qwen3-32B-Instruct2026.06 | 34.8 | |
| GRPOBackbone=Qwen3-32B-Instruct2026.06 | 33.6 | |
| ECHOBackbone=Qwen3-30B-A3B-Instruct2026.06 | 30.5 | |
| SUPOBackbone=Qwen3-30B-A3B-Instruct2026.06 | 26.9 | |
| GRPOBackbone=Qwen3-30B-A3B-Instruct2026.06 | 25.9 |