mixedbread-ai/mxbai-edge-colbert-v0-32m
Primitive: /score · Score ·
ModernBERT
The crispy, lightweight ColBERT family from Mixedbread.
Long context
Overview
Hardware: — drives latency, throughput & cost
| Size | 32M params |
|---|---|
| Tasks | /encode · /score |
| License | apache-2.0 |
| Languages | en |
| 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
default
Quality
map at 10 0.4470
mrr at 10 0.6924
muvera
Quality
ndcg at 10 0.6106
map at 10 0.4525
mrr at 10 0.7072
CMedQAv1-reranking
Chinese medical question answering reranking (v1)
Corpus: 100,000 Queries: 2,000
default
Quality
map at 10 0.2671
mrr at 10 0.3545
muvera
Quality
ndcg at 10 0.2032
map at 10 0.1490
mrr at 10 0.2013
CMedQAv2-reranking
Chinese medical question answering reranking (v2)
Corpus: 108,000 Queries: 4,000
muvera
Quality
ndcg at 10 0.2040
map at 10 0.1521
mrr at 10 0.2038
default
Quality
map at 10 0.2859
mrr at 10 0.3733
CQADupstackPhysicsRetrieval
Duplicate question retrieval from StackExchange Physics
Corpus: 38,314 Queries: 1,039
default_candidates-k-50_candidates-model-Alibaba-NLP__gte-multilingual-base
Quality
map at 10 0.3563
mrr at 10 0.4116
ndcg at 10 0.4086
muvera_candidates-k-50_candidates-model-Alibaba-NLP__gte-multilingual-base
Quality
map at 10 0.1990
mrr at 10 0.2314
ndcg at 10 0.2429
CosQA
Code search with natural language queries
Corpus: 6,267 Queries: 500
default_candidates-k-50_candidates-model-Alibaba-NLP__gte-multilingual-base
Quality
map at 10 0.2056
mrr at 10 0.1992
ndcg at 10 0.2734
muvera_candidates-k-50_candidates-model-Alibaba-NLP__gte-multilingual-base
Quality
map at 10 0.1151
mrr at 10 0.1333
ndcg at 10 0.1500
FiQA2018
Financial opinion mining and question answering
Corpus: 57,599 Queries: 648
default_candidates-k-50_candidates-model-Alibaba-NLP__gte-multilingual-base
Quality
map at 10 0.2762
mrr at 10 0.4134
ndcg at 10 0.3438
muvera_candidates-k-50_candidates-model-Alibaba-NLP__gte-multilingual-base
Quality
map at 10 0.0900
mrr at 10 0.1469
ndcg at 10 0.1303
LegalBenchConsumerContractsQA
Question answering on consumer contracts
Corpus: 153 Queries: 396
default_candidates-k-50_candidates-model-Alibaba-NLP__gte-multilingual-base
Quality
map at 10 0.6869
mrr at 10 0.6872
ndcg at 10 0.7373
muvera_candidates-k-50_candidates-model-Alibaba-NLP__gte-multilingual-base
Quality
map at 10 0.4436
mrr at 10 0.4411
ndcg at 10 0.4993
MMarcoReranking
Multilingual MARCO passage reranking (Chinese)
muvera
Quality
ndcg at 10 0.0862
map at 10 0.0705
mrr at 10 0.0710
default
Quality
ndcg at 10 0.1625
map at 10 0.1352
mrr at 10 0.1352
NFCorpus
Biomedical literature search from NutritionFacts.org
Corpus: 3,593 Queries: 323
default_candidates-k-50_candidates-model-Alibaba-NLP__gte-multilingual-base
Quality
ndcg at 10 0.3557
map at 10 0.1367
mrr at 10 0.5686
muvera_candidates-k-50_candidates-model-Alibaba-NLP__gte-multilingual-base
Quality
map at 10 0.0749
mrr at 10 0.4083
ndcg at 10 0.2290
SCIDOCS
Citation prediction, document classification, and recommendation for scientific papers
Corpus: 25,656 Queries: 1,000
default_candidates-k-50_candidates-model-Alibaba-NLP__gte-multilingual-base
Quality
map at 10 0.0863
mrr at 10 0.2637
ndcg at 10 0.1486
muvera_candidates-k-50_candidates-model-Alibaba-NLP__gte-multilingual-base
Quality
map at 10 0.0602
mrr at 10 0.1965
ndcg at 10 0.1090
SciFact
Scientific claim verification using research literature
Corpus: 5,183 Queries: 300
default_candidates-k-50_candidates-model-Alibaba-NLP__gte-multilingual-base
Quality
map at 10 0.6513
mrr at 10 0.6632
ndcg at 10 0.6960
muvera_candidates-k-50_candidates-model-Alibaba-NLP__gte-multilingual-base
Quality
map at 10 0.3695
mrr at 10 0.3774
ndcg at 10 0.4059
StackOverflowQA
Programming question answering from Stack Overflow
Corpus: 19,931 Queries: 1,994
default_candidates-k-50_candidates-model-Alibaba-NLP__gte-multilingual-base
Quality
map at 10 0.3899
mrr at 10 0.3899
ndcg at 10 0.4215
muvera_candidates-k-50_candidates-model-Alibaba-NLP__gte-multilingual-base
Quality
map at 10 0.0761
mrr at 10 0.0761
ndcg at 10 0.0899
T2Reranking
Chinese passage ranking benchmark
default
Quality
map at 10 0.5354
mrr at 10 0.7578
muvera
Quality
ndcg at 10 0.6687
map at 10 0.4893
mrr at 10 0.7056
Compare (0)Compare →