Reviewer Retrieval on LR-Bench Paper-Centric (test)
0.1698Expertise-aligned LossMERIT-Retriever-8B
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MERIT-Retriever-8BMethod Category=Embedding-based Methods2026.05 | 0.1698 | 75 | |
| RATE-8BMethod Category=Embedding-based Methods2026.05 | 0.1851 | 75 | |
| SPECTER2 PRXMethod Category=Embedding-based Methods2026.05 | 0.1966 | 72.97 | |
| SciNCLMethod Category=Embedding-based Methods2026.05 | 0.2042 | 72.3 | |
| SPECTERMethod Category=Embedding-based Methods2026.05 | 0.2118 | 72.97 | |
| BERTScoreMethod Category=Embedding-based Methods2026.05 | 0.2691 | 66.22 | |
| Qwen3-MaxMethod Category=LLM-based Methods, Decoding Strategy=greedy, Thinking Mode=enabled2026.05 | 0.2698 | 47.65 | |
| DeepSeek-V3.2Method Category=LLM-based Methods, Decoding Strategy=greedy, Thinking Mode=enabled2026.05 | 0.2736 | 50.34 | |
| TPMSMethod Category=Statistical-based Methods2026.05 | 0.292 | 66.22 | |
| CoFMethod Category=Embedding-based Methods2026.05 | 0.2996 | 64.19 |