BAAI/bge-reranker-base
Primitive: /score · Score ·
XLM-RoBERTa
We have updated the new reranker, supporting larger lengths, more languages, and achieving better performance.
Multilingual
Overview
Hardware: — drives latency, throughput & cost
| Size | 278M params |
|---|---|
| Tasks | /score |
| License | mit |
| Languages | en, zh |
| Latency | 45 ms |
| Throughput | 21.3K tok/s |
| Cost | $0.010 /1M tok |
Cost is approximate — computed from list GPU prices; your actual price depends on the provider you deploy SIE with.
Scoring
| Inputs | text |
|---|---|
| Max sequence length | 512 |
Benchmarks
AskUbuntuDupQuestions
Duplicate question detection from AskUbuntu
Corpus: 6,743 Queries: 360
Quality
ndcg at 10 0.5926
map at 10 0.4326
mrr at 10 0.6741
Performance L4 b1 c16
Query 5.0K tok/s
Query p50 33.2ms
CMedQAv1Reranking
Chinese medical question answering reranking (v1)
Corpus: 100,000 Queries: 2,000
Quality
map at 10 0.8073
mrr at 10 0.8414
CMedQAv2Reranking
Chinese medical question answering reranking (v2)
Corpus: 108,000 Queries: 4,000
Quality
map at 10 0.8358
mrr at 10 0.8679
MMarcoReranking
Multilingual MARCO passage reranking (Chinese)
Quality
map at 10 0.3422
mrr at 10 0.3460
Performance L4 b1 c16
Compare (0)Compare →