Question Answering on PopQA (Accuracy, Recall)
62.545AccuracyRAPTOR
Evaluation results
| Method | Links | ||
|---|---|---|---|
| RAPTORLLM=Llama-3-8B, Embedding model=BGE-M3, top-k=4, chunk size=1,200-token, Clustering method=Gaussian Mixture2025.03 | 62.545 | 27.304 | |
| VanillaRAGLLM=Llama-3-8B, Embedding model=BGE-M3, top-k=4, chunk size=1,200-token2025.03 | 60.829 | 27.058 | |
| HippoRAGLLM=Llama-3-8B, Embedding model=BGE-M3, top-k=4, chunk size=1,200-token2025.03 | 59.9 | 24.946 | |
| KGPLLM=Llama-3-8B, Embedding model=BGE-M3, top-k=4, chunk size=1,200-token2025.03 | 57.255 | 24.635 | |
| FastGraphRAGLLM=Llama-3-8B, Embedding model=BGE-M3, top-k=4, chunk size=1,200-token2025.03 | 53.324 | 22.433 | |
| ToGLLM=Llama-3-8B, Embedding model=BGE-M3, top-k=4, chunk size=1,200-token2025.03 | 47.677 | 23.727 | |
| DALKLLM=Llama-3-8B, Embedding model=BGE-M3, top-k=4, chunk size=1,200-token2025.03 | 45.604 | 19.159 | |
| LGraphRAGLLM=Llama-3-8B, Embedding model=BGE-M3, top-k=4, chunk size=1,200-token2025.03 | 45.461 | 18.657 | |
| HLightRAGLLM=Llama-3-8B, Embedding model=BGE-M3, top-k=4, chunk size=1,200-token2025.03 | 41.244 | 18.071 | |
| LLightRAGLLM=Llama-3-8B, Embedding model=BGE-M3, top-k=4, chunk size=1,200-token2025.03 | 38.885 | 16.764 | |
| ZeroShotLLM=Llama-3-8B2025.03 | 28.592 | 8.263 | |
| GLightRAGLLM=Llama-3-8B, Embedding model=BGE-M3, top-k=4, chunk size=1,200-token2025.03 | 20.944 | 8.146 | |
| G-retrieverLLM=Llama-3-8B, Embedding model=BGE-M3, top-k=4, chunk size=1,200-token2025.03 | 17.084 | 6.075 |