lightonai/mLateOn
Primitive: /score · Score ·
ModernBERT
State-of-the-Art Multilingual ColBERT Retrieval Model by LightOn
MultilingualLong context
View on Hugging Face → Fine-tuned from lightonai/mLateOn-unsupervised
Overview
Hardware: — drives latency, throughput & cost
| Size | 307M params |
|---|---|
| Tasks | /encode · /score |
| License | apache-2.0 |
| Languages | en, fr, de, it, es, pt, sv, no, ar, code |
| Latency | — |
| Throughput | — |
| Cost | — /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 | 8,192 |
Benchmarks
AskUbuntuDupQuestions
Duplicate question detection from AskUbuntu
Corpus: 6,743 Queries: 360
Quality
ndcg at 10 0.6627
map at 10 0.5108
mrr at 10 0.7384
CMedQAv1-reranking
Chinese medical question answering reranking (v1)
Corpus: 100,000 Queries: 2,000
Quality
ndcg at 10 0.7572
map at 10 0.6975
mrr at 10 0.7514
CMedQAv2-reranking
Chinese medical question answering reranking (v2)
Corpus: 108,000 Queries: 4,000
Quality
ndcg at 10 0.7646
map at 10 0.7010
mrr at 10 0.7612
MMarcoReranking
Multilingual MARCO passage reranking (Chinese)
Quality
ndcg at 10 0.3284
map at 10 0.2652
mrr at 10 0.2677
T2Reranking
Chinese passage ranking benchmark
Quality
ndcg at 10 0.7460
map at 10 0.5776
mrr at 10 0.8002
Compare (0)Compare →