Long-context evaluation (Financial) on Loong Fin
58.8Fin Judge ScoreRR+Judge(y)
Evaluation results
| Method | Links | |
|---|---|---|
| RR+Judge(y)Context Type=Full Context, Max Length=1M tokens2026.01 | 58.8 | |
| RAO(y)Context Type=Full Context, Max Length=1M tokens2026.01 | 55.3 | |
| RID+Q(y)Context Type=Full Context, Max Length=1M tokens2026.01 | 51.5 | |
| RID(y)Context Type=Full Context, Max Length=1M tokens2026.01 | 51.2 | |
| BaseContext Type=Full Context, Max Length=1M tokens2026.01 | 49.9 | |
| RAGContext Length=32k2026.01 | 47.8 | |
| RID+C(y)Context Type=Full Context, Max Length=1M tokens2026.01 | 47.5 | |
| RID(y)Context Type=KV-Cache with RetrievalAttn2026.01 | 43 | |
| RAO(y)Context Type=KV-Cache with RetrievalAttn2026.01 | 41.7 | |
| RR+Judge(y)Context Type=KV-Cache with RetrievalAttn2026.01 | 41.5 | |
| RID+C(y)Context Type=KV-Cache with RetrievalAttn2026.01 | 39.8 | |
| RID+Q(y)Context Type=KV-Cache with RetrievalAttn2026.01 | 39.7 | |
| BaseContext Type=KV-Cache with RetrievalAttn2026.01 | 39.5 |