sentence-transformers
ONNX
Safetensors
ColBERT
multi-vector
passage-retrieval
custom_code
🇪🇺 Region: EU
Instructions to use jinaai/jina-colbert-v2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- sentence-transformers
How to use jinaai/jina-colbert-v2 with sentence-transformers:
from sentence_transformers import SentenceTransformer model = SentenceTransformer("jinaai/jina-colbert-v2", trust_remote_code=True) sentences = [ "The weather is lovely today.", "It's so sunny outside!", "He drove to the stadium." ] embeddings = model.encode(sentences) similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] - Notebooks
- Google Colab
- Kaggle
better behavior when query_prefix or document_prefix delete
#21
by chenxiaobin - opened
our document is relatively long , approximately 300 to 500 tokens.
we set "doc_maxlen": 2048 to deal with long text。
After deleting the query_prefix or document_prefix, the performance has improved significantly, with the recall rate increasing from 0.8 to 0.84.
This is not a coincidence—we have conducted extensive testing.
do you have any idea to explain this?