Jina Embeddings v2 Base Code
Jina Embeddings v2 Base Code is a JinaBERT encoder with an 8,192-token input window. It creates pooled vectors for semantic search and similarity, rather than generating an answer.
How the model works
The backbone was pretrained on GitHub source code. Jina then used more than 150 million coding question-answer and docstring-code pairs. Its card lists English and 30 programming or markup languages, including Python, JavaScript, SQL and HTML.Jina Embeddings v2 Base Code model details
Use and access
The Apache 2.0 download uses mean pooling and the JinaBERT implementation. Its extended input length comes from symmetric bidirectional ALiBi. This entry preserves the specific trained checkpoint. Later models with different dimensions or training should not be substituted into an existing vector index without rebuilding compatible representations.Jina Embeddings v2 Base Code model detailsJina Embeddings v2 Base Code technical overview
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | |
| Access | Downloadable |
| Model identifiers | jina-embeddings-v2-base-codejinaai/jina-embeddings-v2-base-code |
| Context window | 8,192 tokens Maximum input sequence in the model card; serving limits can be lower. Jina Embeddings v2 Base Code model details |
| Parameters | 161 millionJina Embeddings v2 Base Code model details |
| Weight license | Apache 2.0Jina Embeddings v2 Base Code model details |
| Base model | JinaBERTJina Embeddings v2 Base Code model details |
Related models
- Jina Code Embeddings 0.5B
Jina AI · Jina Code Embeddings
A compact encoder that finds code from descriptions, examples and completion context.
Launch date API · Downloadable - Jina Code Embeddings 1.5B
Jina AI · Jina Code Embeddings
A larger code-search encoder with adjustable vectors for technical retrieval.
Launch date API · Downloadable