NovaSearch/stella_en_1.5B_v5
We released a Jasper and Stella model technology report and code.(2025.1)
Overview
Benchmarks
CQADupstackPhysicsRetrieval
Duplicate question retrieval from StackExchange Physics
Corpus: 38,314 Queries: 1,039
Quality
ndcg at 10 0.5434
map at 10 0.4768
mrr at 10 0.5332
Performance L4 b1 c16
Corpus 11.7K tok/s
Corpus p50 180.9ms
Query 2.0K tok/s
Query p50 73.7ms
CosQA
Code search with natural language queries
Corpus: 6,267 Queries: 500
Quality
ndcg at 10 0.3457
map at 10 0.2713
mrr at 10 0.2795
Performance L4 b1 c16
Corpus 8.5K tok/s
Corpus p50 100.4ms
Query 1.2K tok/s
Query p50 71.8ms
FiQA2018
Financial opinion mining and question answering
Corpus: 57,599 Queries: 648
Quality
ndcg at 10 0.6074
map at 10 0.5230
mrr at 10 0.6780
Performance L4 b1 c16
Corpus 13.0K tok/s
Corpus p50 222.0ms
Query 2.2K tok/s
Query p50 74.6ms
LegalBenchConsumerContractsQA
Question answering on consumer contracts
Corpus: 153 Queries: 396
Quality
ndcg at 10 0.8198
map at 10 0.7715
mrr at 10 0.7728
Performance L4 b1 c16
Corpus 16.7K tok/s
Corpus p50 522.2ms
Query 2.7K tok/s
Query p50 84.9ms
NFCorpus
Biomedical literature search from NutritionFacts.org
Corpus: 3,593 Queries: 323
Quality
ndcg at 10 0.4219
map at 10 0.1649
mrr at 10 0.6231
Performance L4 b1 c16
Corpus 13.9K tok/s
Corpus p50 185.3ms
Query 849 tok/s
Query p50 51.9ms
NanoFiQA2018Retrieval
Smaller subset of the FiQA financial QA dataset
Quality
ndcg at 10 0.6555
map at 10 0.5859
mrr at 10 0.6764
Performance L4 b1 c16
Corpus 1.2K tok/s
Corpus p50 1.5s
Query 1.1K tok/s
Query p50 117.8ms
SCIDOCS
Citation prediction, document classification, and recommendation for scientific papers
Corpus: 25,656 Queries: 1,000
Quality
ndcg at 10 0.2666
map at 10 0.1649
mrr at 10 0.4293
Performance L4 b1 c16
Corpus 12.5K tok/s
Corpus p50 257.9ms
Query 2.0K tok/s
Query p50 77.3ms
SciFact
Scientific claim verification using research literature
Corpus: 5,183 Queries: 300
Quality
ndcg at 10 0.7994
map at 10 0.7496
mrr at 10 0.7579
Performance L4 b1 c16
Corpus 12.8K tok/s
Corpus p50 356.8ms
Query 3.0K tok/s
Query p50 80.0ms
StackOverflowQA
Programming question answering from Stack Overflow
Corpus: 19,931 Queries: 1,994
Quality
ndcg at 10 0.9205
map at 10 0.9063
mrr at 10 0.9063
Performance L4 b1 c16
Corpus 14.2K tok/s
Corpus p50 265.9ms
Query 15.6K tok/s
Query p50 329.0ms