🐦🔥 I've just published Sentence Transformers v5.2.0! It introduces multi-processing for CrossEncoder (rerankers), multilingual NanoBEIR evaluators, similarity score outputs in mine_hard_negatives, Transformers v5 support and more. Details:
- CrossEncoder multi-processing: Similar to SentenceTransformer and SparseEncoder, you can now use multi-processing with CrossEncoder rerankers. Useful for multi-GPU and CPU settings, and simple to configure: just device=["cuda:0", "cuda:1"] or device=["cpu"]*4 on the model.predict or model.rank calls.
- Multilingual NanoBEIR Support: You can now use community translations of the tiny NanoBEIR retrieval benchmark instead of only the English one, by passing dataset_id, e.g. dataset_id="lightonai/NanoBEIR-de" for the German benchmark.
- Similarity scores in Hard Negatives Mining: When mining for hard negatives to create a strong training dataset, you can now pass output_scores=True to get similarity scores returned. This can be useful for some distillation losses!
- Transformers v5: This release works with both Transformers v4 and the upcoming v5. In the future, Sentence Transformers will only work with Transformers v5, but not yet!
- Python 3.9 deprecation: Now that Python 3.9 has lost security support, Sentence Transformers no longer supports it.
I'm quite excited about what's coming. There's a huge draft PR with a notable refactor in the works that should bring some exciting support. Specifically, better multimodality, rerankers, and perhaps some late interaction in the future!
🤗 Sentence Transformers is joining Hugging Face! 🤗 This formalizes the existing maintenance structure, as I've personally led the project for the past two years on behalf of Hugging Face! Details:
Today, the Ubiquitous Knowledge Processing (UKP) Lab is transferring the project to Hugging Face. Sentence Transformers will remain a community-driven, open-source project, with the same open-source license (Apache 2.0) as before. Contributions from researchers, developers, and enthusiasts are welcome and encouraged. The project will continue to prioritize transparency, collaboration, and broad accessibility.
We see an increasing wish from companies to move from large LLM APIs to local models for better control and privacy, reflected in the library's growth: in just the last 30 days, Sentence Transformer models have been downloaded >270 million times, second only to transformers.
I would like to thank the UKP Lab, and especially Nils Reimers and Iryna Gurevych, both for their dedication to the project and for their trust in myself, both now and two years ago. Back then, neither of you knew me well, yet you trusted me to take the project to new heights. That choice ended up being very valuable for the embedding & Information Retrieval community, and I think this choice of granting Hugging Face stewardship will be similarly successful.
I'm very excited about the future of the project, and for the world of embeddings and retrieval at large!
ModernBERT goes MULTILINGUAL! One of the most requested models I've seen, The Johns Hopkins University's CLSP has trained state-of-the-art massively multilingual encoders using the ModernBERT architecture: mmBERT.
Model details: - 2 model sizes: - jhu-clsp/mmBERT-small - jhu-clsp/mmBERT-base - Uses the ModernBERT architecture, but with the Gemma2 multilingual tokenizer (so: flash attention, alternating global/local attention, unpadding/sequence packing, etc.) - Maximum sequence length of 8192 tokens, on the high end for encoders - Trained on 1833 languages using DCLM, FineWeb2, and many more sources - 3 training phases: 2.3T tokens pretraining on 60 languages, 600B tokens mid-training on 110 languages, and 100B tokens decay training on all 1833 languages. - Both models are MIT Licensed, and the full datasets and intermediary checkpoints are also publicly released
Evaluation details: - Very competitive with ModernBERT at equivalent sizes on English (GLUE, MTEB v2 English after finetuning) - Consistently outperforms equivalently sized models on all Multilingual tasks (XTREME, classification, MTEB v2 Multilingual after finetuning) - In short: beats commonly used multilingual base models like mDistilBERT, XLM-R (multilingual RoBERTa), multilingual MiniLM, etc. - Additionally: the ModernBERT-based mmBERT is much faster than the alternatives due to its architectural benefits. Easily up to 2x throughput in common scenarios.
Based on these results, mmBERT should be the new go-to multilingual encoder base models at 300M and below. Do note that the mmBERT models are "base" models, i.e. they're currently only trained to perform Mask Filling. They'll need to be finetuned for downstream tasks like semantic search, classification, clustering, etc.
Supercharge Apple’s Shortcuts using Cloudflare Workers and Gemini within minutes (and for free, up to 1,500 requests per day) ☁️✨
Hello everyone, last week, while experimenting for fun, I created an API that allows you to easily access AI models (in this case, Google's) from the Shortcut app in order to analyze data from my apps and make the most of it thanks to the generative capabilities of advanced models.
It costs me nothing, and I think it might be good to share it so that others can build on it.
In README.md, you will find everything you need to get started and put your own microservice into production, which you can call from the app’s HTTP request features.
You will simply be asked to have a free Cloudflare account and an API key obtained from Google's AI Studio.
Feel free to take a look and get back to me if you encounter any problems during deployment.
Although more and more code editors are aligning themselves with the AGENTS.md file standard, some still use specific nomenclatures that can make it difficult to maintain different configuration files when several people are working on the same project with different agents.
Bodyboard addresses this by generating canonical instructions for code helpers from a single AGENTS.md file, thereby streamlining the production of adapter outputs for Gemini CLI, Copilot, Cline, Claude, Rules, Windsurf, and OpenAI Codex integrations.
😎 I just published Sentence Transformers v5.1.0, and it's a big one. 2x-3x speedups of SparseEncoder models via ONNX and/or OpenVINO backends, easier distillation data preparation with hard negatives mining, and more:
1️⃣ Faster ONNX and OpenVINO backends for SparseEncoder models Usage is as simple as backend="onnx" or backend="openvino" when initializing a SparseEncoder to get started, but I also included utility functions for optimization, dynamic quantization, and static quantization, plus benchmarks.
2️⃣ New n-tuple-scores output format from mine_hard_negatives This new output format is immediately compatible with the MarginMSELoss and SparseMarginMSELoss for training SentenceTransformer, CrossEncoder, and SparseEncoder losses.
3️⃣ Gathering across devices When doing multi-GPU training using a loss that has in-batch negatives (e.g. MultipleNegativesRankingLoss), you can now use gather_across_devices=True to load in-batch negatives from the other devices too! Essentially a free lunch, pretty big impact potential in my evals.
4️⃣ Trackio support If you also upgrade transformers, and you install trackio with pip install trackio, then your experiments will also automatically be tracked locally with trackio. Just open up localhost and have a look at your losses/evals, no logins, no metric uploading.
5️⃣ MTEB Documentation We've added some documentation on evaluating SentenceTransformer models properly with MTEB. It's rudimentary as the documentation on the MTEB side is already great, but it should get you started.
Plus many more smaller features & fixes (crash fixes, compatibility with datasets v4, FIPS compatibility, etc.).
Big thanks to all of the contributors for helping with the release, many of the features from this release were proposed by others. I have a big list of future potential features that I'd love to add, but I'm
Qwen 3 Coder is a personal attack to k2, and I love it. It achieves near SOTA on LCB while not having reasoning. Finally people are understanding that reasoning isnt necessary for high benches...
Because hackathons are often the starting point for many AI projects, I've created a Python-backend template incorporating my feedback to streamline collaboration and urgent deployments 🏎️
Within a year, I had the opportunity to participate in hackathons organized by Mistral, OpenAI, and DeepMind and this GitHub template is structured around several fundamental building blocks and recommendations I offer developers eager to participate in their first hackathon, whether as part of a team or individually. Its emphasis is on rapid setup and deployment through: - uv as a package manager, simplifying usage via a series of pre-configured make commands. - FastAPI for API management, structured in a modular architecture designed to minimize branch conflicts during merges to main branches (using minimal health-check and ping routes to verify Docker’s proper execution and backend accessibility on the local network). - Pydantic for validation and type handling, which simplifies debugging and enhances understanding of data objects. - A set of custom instructions tailored for agents (Cline and GitHub Copilot), aimed at improving overall comprehension of the application and optimizing the vibe-coding experience.
This template includes unit tests with a 100% success rate and test coverage, as well as a minimal CI file ensuring that the FastAPI application runs correctly. Thus, merging code that breaks the server into production becomes impossible ⛔️
In general, I would reiterate an essential piece of advice: your two main adversaries are branch conflicts—particularly when the same file is modified concurrently within a brief period, especially if your architecture isn’t built for scalability—and deployment issues under urgent circumstances ⏱️
‼️Sentence Transformers v5.0 is out! The biggest update yet introduces Sparse Embedding models, encode methods improvements, Router module for asymmetric models & much more. Sparse + Dense = 🔥 hybrid search performance! Details:
1️⃣ Sparse Encoder Models Brand new support for sparse embedding models that generate high-dimensional embeddings (30,000+ dims) where <1% are non-zero:
- Full SPLADE, Inference-free SPLADE, and CSR architecture support - 4 new modules, 12 new losses, 9 new evaluators - Integration with @elastic-co, @opensearch-project, @NAVER LABS Europe, @qdrant, @IBM, etc. - Decode interpretable embeddings to understand token importance - Hybrid search integration to get the best of both worlds
2️⃣ Enhanced Encode Methods & Multi-Processing - Introduce encode_query & encode_document automatically use predefined prompts - No more manual pool management - just pass device list directly to encode() - Much cleaner and easier to use than the old multi-process approach
3️⃣ Router Module & Advanced Training - Router module with different processing paths for queries vs documents - Custom learning rates for different parameter groups - Composite loss logging - see individual loss components - Perfect for two-tower architectures
4️⃣ Comprehensive Documentation & Training - New Training Overview, Loss Overview, API Reference docs - 6 new training example documentation pages - Full integration examples with major search engines - Extensive blogpost on training sparse models
What's next? We would love to hear from the community! What sparse encoder models would you like to see? And what new capabilities should Sentence Transformers handle - multimodal embeddings, late interaction models, or something else? Your feedback shapes our roadmap!
AMD summer hackathons are here! A chance to get hands-on with MI300X GPUs and accelerate models. 🇫🇷 Paris - Station F - July 5-6 🇮🇳 Mumbai - July 12-13 🇮🇳 Bengaluru - July 19-20
Hugging Face and GPU Mode will be on site and on July 6 in Paris @ror will share lessons learned while building new kernels to accelerate Llama 3.1 405B on ROCm
🌐 Clinical Trials Dataset now available on Hugging Face! 🧬
I’ve just released a comprehensive, ML-ready dataset featuring 500,000+ clinical trial records sourced directly from ClinicalTrials.gov for biomedical NLP, healthcare analytics, and clinical research applications 🤗
I wanted to produce the most complete and up-to-date dump with all raw data partially flattened to simplify extraction, self-querying and processing.
Do you have any ideas about what we can do with it? Using descriptions to enhance specialized embedding models?
It's just become easier to share your apps on the biggest AI app store (aka HF spaces) for unlimited storage, more visibility and community interactions.
Just pick a React, Svelte, or Vue template when you create your space or add app_build_command: npm run build in your README's YAML and app_file: build/index.html in your README's YAML block.
Wrapping up a week of shipping and announcements with Dell Enterprise Hub now featuring AI Applications, on-device models for AI PCs, a new CLI and Python SDK... all you need for building AI on premises!
