Question Answering on SQuAD (Correlation Metrics)
2,075Correlation (alpha=0.17)LLM-RAG
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| LLM-RAGLLM=Qwen3-4B2026.05 | 2,075 | 2,069 | 2,075 | 2,075 | 2,075 | 2,075 | 2,075 | 2,075 | 2,075 | |
| BalanceRAGLLM=Qwen3-4B2026.05 | 2,032 | 2,053 | 2,050 | 2,031 | 2,032 | 2,032 | 2,032 | 2,032 | 2,032 | |
| UCB-Cascaded-HFDLLM=Qwen3-4B2026.05 | 2,019 | — | — | — | — | — | — | — | 2,015 | |
| UCB-Cascaded-CPLLM=Qwen3-4B2026.05 | 2,008 | — | — | — | — | — | 2,019 | 2,018 | 1,983 | |
| LLM-RAGLLM=Qwen3-14B2026.05 | 1,973 | 1,953 | 1,972 | 1,973 | 1,973 | 1,973 | 1,973 | 1,973 | 1,973 | |
| LLM-RAGLLM=Qwen2.5-3B2026.05 | 1,942 | 1,450 | 1,653 | 1,785 | 1,873 | 1,929 | 1,942 | 1,942 | 1,942 | |
| BalanceRAGLLM=Qwen2.5-3B2026.05 | 1,930 | 1,357 | 1,464 | 1,649 | 1,798 | 1,887 | 1,922 | 1,929 | 1,930 | |
| LLM-RAGLLM=Qwen2.5-7B2026.05 | 1,929 | 1,723 | 1,836 | 1,896 | 1,924 | 1,929 | 1,929 | 1,929 | 1,929 | |
| UCB-Cascaded-CPLLM=Qwen2.5-3B2026.05 | 1,928 | — | — | — | — | — | 1,916 | 1,927 | 1,925 | |
| BalanceRAGLLM=Qwen3-14B2026.05 | 1,924 | 1,942 | 1,948 | 1,905 | 1,914 | 1,923 | 1,924 | 1,924 | 1,924 | |
| BalanceRAGLLM=Qwen2.5-7B2026.05 | 1,916 | 1,611 | 1,698 | 1,821 | 1,895 | 1,915 | 1,916 | 1,916 | 1,916 | |
| UCB-Cascaded-HFDLLM=Qwen2.5-7B2026.05 | 1,914 | — | — | — | — | — | — | — | 1,914 | |
| UCB-Cascaded-HFDLLM=Qwen3-14B2026.05 | 1,910 | — | — | — | — | — | — | — | 1,895 | |
| UCB-Cascaded-CPLLM=Qwen2.5-7B2026.05 | 1,906 | — | — | — | — | 1,913 | 1,915 | 1,913 | 1,890 | |
| UCB-Cascaded-CPLLM=Qwen3-14B2026.05 | 1,882 | — | — | — | — | — | 1,912 | 1,903 | 1,853 | |
| LLM-RAGLLM=Llama3.1-8B2026.05 | 1,853 | 1,763 | 1,803 | 1,831 | 1,850 | 1,853 | 1,853 | 1,853 | 1,853 | |
| BalanceRAGLLM=Llama3.1-8B2026.05 | 1,843 | 1,723 | 1,784 | 1,819 | 1,836 | 1,842 | 1,843 | 1,843 | 1,843 | |
| UCB-Cascaded-CPLLM=Llama3.1-8B2026.05 | 1,834 | — | — | — | — | 1,835 | 1,837 | 1,835 | 1,830 | |
| LLM-RAGLLM=Llama3.2-3B2026.05 | 1,809 | 1,438 | 1,546 | 1,617 | 1,666 | 1,714 | 1,755 | 1,784 | 1,829 | |
| BalanceRAGLLM=Llama3.2-3B2026.05 | 1,805 | 1,377 | 1,484 | 1,591 | 1,654 | 1,699 | 1,745 | 1,781 | 1,828 | |
| UCB-Cascaded-CPLLM=Llama3.2-3B2026.05 | 1,798 | — | — | — | — | — | 1,744 | 1,775 | 1,818 | |
| LLM-RAGLLM=Vicuna-7B-v1.52026.05 | 1,484 | 1,330 | 1,376 | 1,411 | 1,443 | 1,464 | 1,479 | 1,484 | 1,484 | |
| BalanceRAGLLM=Vicuna-7B-v1.52026.05 | 1,482 | 1,258 | 1,327 | 1,374 | 1,415 | 1,447 | 1,468 | 1,479 | 1,482 | |
| UCB-Cascaded-CPLLM=Vicuna-7B-v1.52026.05 | 1,473 | — | — | — | — | — | — | 1,467 | 1,473 | |
| LLM-RAGLLM=Vicuna-13B-v1.52026.05 | 1,291 | 1,027 | 1,094 | 1,143 | 1,183 | 1,224 | 1,256 | 1,280 | 1,292 | |
| BalanceRAGLLM=Vicuna-13B-v1.52026.05 | 1,276 | 950 | 1,018 | 1,097 | 1,149 | 1,189 | 1,224 | 1,256 | 1,281 | |
| UCB-Cascaded-CPLLM=Vicuna-13B-v1.52026.05 | 1,266 | — | — | — | — | — | — | 1,245 | 1,279 | |
| LLM-onlyLLM=Qwen3-14B2026.05 | 1,251 | — | — | — | — | — | 974 | 1,094 | 1,386 | |
| LLM-onlyLLM=Qwen3-4B2026.05 | 948 | — | — | — | — | — | 781 | 838 | 1,120 | |
| LLM-onlyLLM=Qwen2.5-7B2026.05 | 682 | — | — | — | — | 443 | 468 | 551 | 878 | |
| LLM-onlyLLM=Qwen2.5-3B2026.05 | 402 | — | — | — | — | — | 333 | 362 | 452 | |
| LLM-onlyLLM=Llama3.1-8B2026.05 | 257 | — | — | — | — | 194 | 211 | 227 | 288 | |
| LLM-onlyLLM=Llama3.2-3B2026.05 | 164 | — | — | — | — | — | 137 | 144 | 211 | |
| LLM-onlyLLM=Vicuna-7B-v1.52026.05 | 138 | — | — | — | — | — | — | 130 | 150 | |
| LLM-onlyLLM=Vicuna-13B-v1.52026.05 | 136 | — | — | — | — | — | — | 128 | 152 |