Tom Aarsen

Or check out the models & datasets directly via this Collection: huggingface.co/collections/...

Ettin Rerankers - a cross-encoder Collection huggingface.co

Read the full blog post for the model links, results, recipe, and the ~150 line training script. Or just point your Agent at the URL: huggingface.co/blog/ettin-r...

Introducing the Ettin Reranker Family huggingface.co

Full release notes: github.com/huggingface/... pip install sentence-transformers==5.6.1

Release v5.6.1 - Flash Attention Fix for XLM-R and RoBERTa Models · huggingface/sentence-transformers github.com

Or the models: huggingface.co/collections/...

LightOn-rerank - a lightonai Collection huggingface.co

Check out the full blogpost with a ton of information: huggingface.co/blog/lighton...

One Adapter, Both Modalities: Field Notes from Building and Serving a Multimodal Reranker huggingface.co

Or check out the models directly here: huggingface.co/collections/...

Nemotron 3 Embed - a nvidia Collection huggingface.co

Read all the details in their announcement blogpost: huggingface.co/blog/nvidia/...

NVIDIA Nemotron 3 Embed Ranks #1 Overall on RTEB, Advancing Agentic Retrieval huggingface.co

The reranker: huggingface.co/tencent/R3-r...

tencent/R3-rerank-0.6b · Hugging Face huggingface.co

The embedding model: huggingface.co/tencent/R3-e...

tencent/R3-embedding-0.6b · Hugging Face huggingface.co

The free space, no login needed: huggingface.co/spaces/huggi...

V-SPLADE Quality Document Retrieval - a Hugging Face Space by hugging-apps huggingface.co

Efficient: huggingface.co/naver/v-spla... 🧵

naver/v-splade-efficient · Hugging Face huggingface.co

Quality: huggingface.co/naver/v-spla... 🧵

naver/v-splade-quality · Hugging Face huggingface.co

Tencent just published R3-Skill, a two-stage retrieval stack purpose-built for a problem RAG-style retrievers weren't designed for: routing LLM agent skills (think Anthropic's SKILLmd format). Two 0.6B models, both Apache 2.0, one embedding model, and one reranker. 🧵

🎉 @lightonai.bsky.social just published LightOn-rerank: rerankers that score text passages or document page images against a query. Six models: Qwen3.5 at 0.8B / 2B / 4B, each in a pointwise and a generative listwise variant. Excellent for text <-> image retrieval. 🧵

💧 Liquid AI released 2 multilingual retrieval models, the first bidirectional members of the LFM family. Both 350M params, 11 languages (ar, de, en, es, fr, it, ja, ko, no, pt, sv): - LFM2.5-Embedding-350M (bi-encoder) - LFM2.5-ColBERT-350M (multi-vector, late interaction) 🧵