List your company

Question Answering on HotpotQA (test) (EM, F1, TTFT, E2E, Ret/1K, Eff., QAL Metrics)

70.3EMOracle
31.341.42551.5561.675May 18, 2026

Evaluation results

MethodLinks
OracleBackbone=Llama-3.1-8B, Retrieval Infrastructure=Wikipedia corpus, FAISS IVF index, Contriever embeddings2026.0570.376.2453.14824.64.4
Sync-RAGBackbone=Llama-3.1-8B, Retrieval Infrastructure=Wikipedia corpus, FAISS IVF index, Contriever embeddings2026.0569.275.12879.2868.213.3
Ours (Predictive)Backbone=Llama-3.1-8B, Retrieval Infrastructure=Wikipedia corpus, FAISS IVF index, Contriever embeddings2026.0568.775.11085.25914.47.6
Adaptive-RAGBackbone=Llama-3.1-8B, Retrieval Infrastructure=Wikipedia corpus, FAISS IVF index, Contriever embeddings2026.0568.574.32157.16810.510.4
Self-RAGBackbone=Llama-3.1-8B, Retrieval Infrastructure=Wikipedia corpus, FAISS IVF index, Contriever embeddings2026.0567.873.62347.8729.411.5
DRAGINBackbone=Llama-3.1-8B, Retrieval Infrastructure=Wikipedia corpus, FAISS IVF index, Contriever embeddings2026.0567.373.21786.46411.49.5
PipeRAGBackbone=Llama-3.1-8B, Retrieval Infrastructure=Wikipedia corpus, FAISS IVF index, Contriever embeddings, Retrieval Interval=322026.0566.872.91185.666.8138.4
FLAREBackbone=Llama-3.1-8B, Retrieval Infrastructure=Wikipedia corpus, FAISS IVF index, Contriever embeddings2026.0565.972.12066.87110.610.3
Entropy-ThresholdBackbone=Llama-3.1-8B, Retrieval Infrastructure=Wikipedia corpus, FAISS IVF index, Contriever embeddings2026.0564.270.52688.7828.113.6
No-RAGBackbone=Llama-3.1-8B2026.0532.839.4422.3017.17