#jinaai — Public Fediverse posts

long US weekend — great time to catch up on some #jinaAI papers about rerankers and code / multilingual / multimodal embeddings:
* jina-reranker-v3
* jina-code-embeddings
* jina-embeddings-v4
* ReaderLM-v2
* jina-clip-v2

I spy a #JinaAI model on elastic cloud 🙃

Jina-VLM-2.4B đạt SOTA cho câu hỏi đa ngôn ngữ với encoder SigLIP2 và decoder Qwen3, cho điểm trung bình 72.3 trên 8 benchmark VQA, cùng 78.8 (MMMB) & 74.3 (Multilingual MMBench). Đào tạo từ 5M dữ liệu đa phương tiện & 12B tokens 29 ngôn ngữ. #AI #MachineLearning #JinaAI #TienTien #TríTuNhanTạo

*(Translated and summarized with key stats, no URLs included)*

https://www.reddit.com/r/LocalLLaMA/comments/1ph9pg9/new_jinavlm24b_reaches_sota_for_multilingual/

kicking off #ElasticON NYC. I‘ll put the most relevant context (😉) in this thread — it will be a busy day
starting with joining forces with #JinaAI https://www.elastic.co/blog/elastic-jina-ai 🤗

DeepSearchの精度改善手法：URL Rankingについて
https://qiita.com/xxyc/items/094f889879ce8e9ec7e0?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items

#qiita #ChatGPT #Rerank #DeepResearch #DeepSearch #JinaAI

better context seems to be a popular topic right now. "Late Chunking in Long-Context Embedding Models" from #JinaAI works very differently but seems to tackle a similar problem: https://jina.ai/news/late-chunking-in-long-context-embedding-models/
though the results there look less clear