The Who's Who of AI

“Very nice work by the NLP & AI Lab at Korea University! 👏 A full Korean-English retrieval toolkit, from three ways to index your documents to reranking your candidates, all usable through Sentence Transformers. huggingface.co/collections/...”

“August's KURE-v2 completes the set with MultiVectorEncoder support. At 154M parameters, it keeps a 128d vector per token and uses MaxSim, letting different query tokens match different parts of a document. huggingface.co/nlpai-lab/KU...”

“KURE-v3 gives you one 2,048d vector per text, with Matryoshka support down to 256d. In the team's Korean retrieval evaluation, 1,024d retains 99% of its full-size score, so you can halve vector storage with very little loss. huggingface.co/nlpai-lab/KU...”

“The collection has Embed 1B, Embed 8B, and Rerank 8B in BF16, plus an NVFP4 version of Embed 1B for Blackwell GPUs. The BF16 models have Sentence Transformers examples in their cards, and all use OpenMDW-1.1: huggingface.co/collections/...”

“Thanks to everyone who contributed to this release! Full release notes, examples & all 176(!) PRs: github.com/huggingface/... pip install sentence-transformers==6.2.0”

“mSlideOn: huggingface.co/lightonai/mS...”

“Nice work to @SimonBrandeis , this is very important work for speeding up some of the most popular Sentence Transformer models, for example. And Static embedding models (huggingface.co/blog/static-...) should be very greatly sped up as well!”

“It also covers a nice demo where you can measure the tokenization speed difference in your browser, and see the difference in tokenization steps before and after the refactor. Here's a direct link: huggingface.co/spaces/sbran...”

“All of the details are in this blog, which explains exactly how these tokenizers work, where the implementation was bottlenecked, and how it has been improved. Check it out here: huggingface.co/blog/hugging...”

“The model: huggingface.co/google/embed... Give it a like if you haven't already, or even better: give it a try, it's natively integrated in Sentence Transformers for any modality you want.”

“And for plotting the Like histories, I used this excellent Space: huggingface.co/spaces/timqi...”

“I quite like github.com/lightonai/fa...! There's some docs here: huggingface.co/blog/multi-v... And indeed, FAISS won't work, as it's not designed for MaxSim and multi-vector embeddings indeed.”

“I quite like github.com/lightonai/fa...! There's some docs here: huggingface.co/blog/multi-v... And indeed, FAISS won't work, as it's not designed for MaxSim and multi-vector embeddings indeed.”

“Credit to the Perplexity Research team for the models. Both are MIT licensed. pip install "sentence-transformers>=6.0.0" "transformers>=5.4.0" huggingface.co/collections/...”

“Check out the models here: huggingface.co/collections/... Or read Perplexity's blogpost: www.perplexity.ai/hub/blog/mul...”

“I think it's ~3 main factors: 1) people are excited about its multimodal possibilities that they don't really have with their current models, 2) it had good release timing, causing it to become the #1 trending model on hf.co/models which resulted in more ey…”