Conversational AI on EverydayGPT
22.6F1 ScoreCGRAG
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CGRAGLatency=6.00 s2026.04 | 22.6 | 19.5 | |
| RAG-OnlyConfidence Threshold (tau)=1.0, Latency=6.29 s2026.04 | 22.4 | 19.3 | |
| GPT-Dom.Confidence Threshold (tau)=0.1, Latency=5.62 s2026.04 | 21.3 | 18.3 | |
| LangChain RAGLatency=0.61 s2026.04 | 21 | 18.1 | |
| BM25+Dense hybridLatency=0.55 s2026.04 | 20.7 | 17.9 | |
| FAISS dense (uncond.)Latency=0.48 s2026.04 | 19.8 | 17.2 | |
| BM25Latency=0.21 s2026.04 | 18.9 | 16.2 | |
| GPT-onlyLatency=5.62 s2026.04 | 17.1 | 14.8 | |
| GPT-2 SmallModel Parameters=117M, Latency=4.80 s2026.04 | 15.8 | 13.6 |