Review Generation on LongLaMP Rev
34.81R1 ScoreSpecSteer
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SpecSteerPair=C, Backbone=Qwen3 8B/32B2026.03 | 34.81 | 16.29 | |
| SpecSteerPair=A, Backbone=Qwen3 0.6B/32B2026.03 | 33.03 | 17.26 | |
| SLM+Pair=C, Backbone=Qwen3 8B/32B, Note=retrieval-augmented specialist2026.03 | 32.77 | 15.42 | |
| SpecSteerPair=B, Backbone=Qwen2.5 1.5B/32B2026.03 | 32.7 | 15.44 | |
| SpecSteerPair=D, Backbone=Llama2026.03 | 31.88 | 14.68 | |
| LLMPair=B, Backbone=Qwen2.5 1.5B/32B2026.03 | 31.73 | 14.41 | |
| LLMPair=D, Backbone=Llama2026.03 | 31.71 | 15.08 | |
| LLMPair=A, Backbone=Qwen3 0.6B/32B2026.03 | 31.18 | 14.78 | |
| LLMPair=C, Backbone=Qwen3 8B/32B2026.03 | 31.18 | 14.78 | |
| SLM+Pair=D, Backbone=Llama, Note=retrieval-augmented specialist2026.03 | 30.53 | 14.52 | |
| SLMPair=C, Backbone=Qwen3 8B/32B2026.03 | 30.43 | 14.54 | |
| SLMPair=D, Backbone=Llama2026.03 | 28.72 | 13.94 | |
| SLM+Pair=B, Backbone=Qwen2.5 1.5B/32B, Note=retrieval-augmented specialist2026.03 | 26.39 | 12.12 | |
| SLMPair=A, Backbone=Qwen3 0.6B/32B2026.03 | 24.15 | 12.95 | |
| SLM+Pair=A, Backbone=Qwen3 0.6B/32B, Note=retrieval-augmented specialist2026.03 | 23.18 | 12.84 | |
| SLMPair=B, Backbone=Qwen2.5 1.5B/32B2026.03 | 22.5 | 10.86 |