List your company

Question Answering on SQuAD v2 (Corr., Err.)

1,853CorrelationLLM-RAG
68.36531.689951,458.32May 19, 2026

Evaluation results

MethodLinks
LLM-RAGBackbone=Llama3.1-8B, Target risk level (alpha)=0.142026.051,85311.93
LLM-RAGBackbone=Llama3.1-8B, Target risk level (alpha)=0.152026.051,85311.93
LLM-RAGBackbone=Llama3.1-8B, Target risk level (alpha)=0.132026.051,85011.75
BalanceRAGBackbone=Llama3.1-8B, Target risk level (alpha)=0.152026.051,84312.4
BalanceRAGBackbone=Llama3.1-8B, Target risk level (alpha)=0.142026.051,84212.41
UCB-CascadedBackbone=Llama3.1-8B, Target risk level (alpha)=0.152026.051,83712.69
BalanceRAGBackbone=Llama3.1-8B, Target risk level (alpha)=0.132026.051,83612.1
UCB-CascadedBackbone=Llama3.1-8B, Target risk level (alpha)=0.142026.051,83512.63
LLM-RAGBackbone=Llama3.1-8B, Target risk level (alpha)=0.122026.051,83110.95
BalanceRAGBackbone=Llama3.1-8B, Target risk level (alpha)=0.122026.051,81911.18
LLM-RAGBackbone=Llama3.1-8B, Target risk level (alpha)=0.112026.051,80310
BalanceRAGBackbone=Llama3.1-8B, Target risk level (alpha)=0.112026.051,78410.1
LLM-RAGBackbone=Llama3.2-3B, Target risk level (alpha)=0.152026.051,75514.16
BalanceRAGBackbone=Llama3.2-3B, Target risk level (alpha)=0.152026.051,74513.75
UCB-CascadedBackbone=Llama3.2-3B, Target risk level (alpha)=0.152026.051,74413.54
LLM-RAGBackbone=Llama3.2-3B, Target risk level (alpha)=0.142026.051,71413.16
BalanceRAGBackbone=Llama3.2-3B, Target risk level (alpha)=0.142026.051,69912.71
LLM-RAGBackbone=Llama3.2-3B, Target risk level (alpha)=0.132026.051,66612.16
BalanceRAGBackbone=Llama3.2-3B, Target risk level (alpha)=0.132026.051,65411.64
LLM-RAGBackbone=Llama3.2-3B, Target risk level (alpha)=0.122026.051,61711.12
BalanceRAGBackbone=Llama3.2-3B, Target risk level (alpha)=0.122026.051,59110.66
LLM-RAGBackbone=Llama3.2-3B, Target risk level (alpha)=0.112026.051,54610.08
BalanceRAGBackbone=Llama3.2-3B, Target risk level (alpha)=0.112026.051,4849.75
LLM-onlyBackbone=Llama3.1-8B, Target risk level (alpha)=0.152026.0521113.24
LLM-onlyBackbone=Llama3.1-8B, Target risk level (alpha)=0.142026.0519413.19
LLM-onlyBackbone=Llama3.2-3B, Target risk level (alpha)=0.152026.0513713.45