Table 4 Mean reciprocal rank (MRR) and mean average precision (MAP) of different systems in our test cases.
From: Dual retrieving and ranking medical large language model with retrieval augmented generation
System components | Mean reciprocal rank | Mean average precision |
|---|---|---|
IvyGPT + Chroma | 0.62 | 0.54 |
IvyGPT + Chroma + ColBERTv2 | 0.66 | 0.59 |
IvyGPT + Elasticsearch + ColBERTv2 | 0.57 | 0.51 |
SELF-RAG | 0.69 | 0.60 |
Our pipeline | 0.72 | 0.63 |
