Update README.md
Browse files
README.md
CHANGED
@@ -304,14 +304,29 @@ Additionally, we provide the following embedding models, you can also use them f
|
|
304 |
Join our [Discord community](https://discord.jina.ai) and chat with other community members about ideas.
|
305 |
|
306 |
```
|
307 |
-
@
|
308 |
-
|
309 |
-
|
310 |
-
|
311 |
-
|
312 |
-
|
313 |
-
|
314 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
315 |
}
|
316 |
```
|
317 |
|
|
|
304 |
Join our [Discord community](https://discord.jina.ai) and chat with other community members about ideas.
|
305 |
|
306 |
```
|
307 |
+
@inproceedings{xiao-etal-2024-jina,
|
308 |
+
title = "{J}ina-{C}ol{BERT}-v2: A General-Purpose Multilingual Late Interaction Retriever",
|
309 |
+
author = {Jha, Rohan and
|
310 |
+
Wang, Bo and
|
311 |
+
G{\"u}nther, Michael and
|
312 |
+
Mastrapas, Georgios and
|
313 |
+
Sturua, Saba and
|
314 |
+
Mohr, Isabelle and
|
315 |
+
Koukounas, Andreas and
|
316 |
+
Wang, Mohammad Kalim and
|
317 |
+
Wang, Nan and
|
318 |
+
Xiao, Han},
|
319 |
+
editor = {S{\"a}lev{\"a}, Jonne and
|
320 |
+
Owodunni, Abraham},
|
321 |
+
booktitle = "Proceedings of the Fourth Workshop on Multilingual Representation Learning (MRL 2024)",
|
322 |
+
month = nov,
|
323 |
+
year = "2024",
|
324 |
+
address = "Miami, Florida, USA",
|
325 |
+
publisher = "Association for Computational Linguistics",
|
326 |
+
url = "https://aclanthology.org/2024.mrl-1.11/",
|
327 |
+
doi = "10.18653/v1/2024.mrl-1.11",
|
328 |
+
pages = "159--166",
|
329 |
+
abstract = "Multi-vector dense models, such as ColBERT, have proven highly effective in information retrieval. ColBERT`s late interaction scoring approximates the joint query-document attention seen in cross-encoders while maintaining inference efficiency closer to traditional dense retrieval models, thanks to its bi-encoder architecture and recent optimizations in indexing and search. In this paper, we introduce a novel architecture and a training framework to support long context window and multilingual retrieval. Leveraging Matryoshka Representation Loss, we further demonstrate that the reducing the embedding dimensionality from 128 to 64 has insignificant impact on the model`s retrieval performance and cut storage requirements by up to 50{\%}. Our new model, Jina-ColBERT-v2, demonstrates strong performance across a range of English and multilingual retrieval tasks,"
|
330 |
}
|
331 |
```
|
332 |
|