I just released Sentence Transformers v3.4.0, featuring a memory leak fix, compatibility between the powerful Cached... losses and the Matryoshka loss modifier, and a bunch of fixes & small features.
🪆 Matryoshka & Cached loss compatibility It is now possible to combine the powerful Cached... losses (which use in-batch negatives & a caching mechanism to allow for endless batch size & negatives) with the Matryoshka loss modifier which modifies a base loss such that it is trained not only on the maximum dimensionality (e.g. 1024 dimensions), but also on many lower dimensions (e.g. 768, 512, 256, 128, 64, 32). After training, these models' embeddings can be truncated for faster retrieval, etc.
🎞️ Resolve memory leak when Model and Trainer are reinitialized Due to a circular dependency between Trainer -> Model -> ModelCardData -> Trainer, deleting both the trainer & model still didn't free up the memory. This led to a memory leak in scripts where you repeatedly do so.
➕ New Features Many new small features, e.g. multi-GPU support for 'mine_hard_negatives', a 'margin' parameter to TripletEvaluator, and Matthews Correlation Coefficient in the BinaryClassificationEvaluator.
🐛 Bug Fixes Also a bunch of fixes, for example that subsequent batches were not sorted when using the "no_duplicates" batch sampler. See the release notes for more details.
Interact with your PDF documents like never before! 🤯 Extract text & images, then ask context-aware questions based on both. Powered by RAG techniques & multimodal LLMs. Perfect for studying, research & more! 📝👀 Try it out now!!!! ✍️
🏎️ Today I'm introducing a method to train static embedding models that run 100x to 400x faster on CPU than common embedding models, while retaining 85%+ of the quality! Including 2 fully open models: training scripts, datasets, metrics.
We apply our recipe to train 2 Static Embedding models that we release today! We release: 2️⃣ an English Retrieval model and a general-purpose Multilingual similarity model (e.g. classification, clustering, etc.), both Apache 2.0 🧠 my modern training strategy: ideation -> dataset choice -> implementation -> evaluation 📜 my training scripts, using the Sentence Transformers library 📊 my Weights & Biases reports with losses & metrics 📕 my list of 30 training and 13 evaluation datasets
The 2 Static Embedding models have the following properties: 🏎️ Extremely fast, e.g. 107500 sentences per second on a consumer CPU, compared to 270 for 'all-mpnet-base-v2' and 56 for 'gte-large-en-v1.5' 0️⃣ Zero active parameters: No Transformer blocks, no attention, not even a matrix multiplication. Super speed! 📏 No maximum sequence length! Embed texts at any length (note: longer texts may embed worse) 📐 Linear instead of exponential complexity: 2x longer text takes 2x longer, instead of 2.5x or more. 🪆 Matryoshka support: allow you to truncate embeddings with minimal performance loss (e.g. 4x smaller with a 0.56% perf. decrease for English Similarity tasks)
Check out the full blogpost if you'd like to 1) use these lightning-fast models or 2) learn how to train them with consumer-level hardware: https://huggingface.co/blog/static-embeddings
The blogpost contains a lengthy list of possible advancements; I'm very confident that our 2 models are only the tip of the iceberg, and we may be able to get even better performance.
Checkout phi-4 from Microsoft, dropped a day ago... If you ❤️ the Phi series, then here is the GGUF - Sri-Vigneshwar-DJ/phi-4-GGUF. phi-4 is a 14B highly efficient open LLM that beats much larger models at math and reasoning - check out evaluations on the Open LLM.
Just sharing a thought: I started using DeepSeek V3 a lot, and an idea struck me about agents "orchestrating during inference" on a test-time compute model like DeepSeek V3 or the O1 series.
Agents (Instruction + Function Calls + Memory) execute during inference, and based on the output decision, a decision is made to scale the time to reason or perform other tasks.
Combining smolagents with Anthropic’s best practices simplifies building powerful AI agents:
1. Code-Based Agents: Write actions as Python code, reducing steps by 30%. 2. Prompt Chaining: Break tasks into sequential subtasks with validation gates. 3. Routing: Classify inputs and direct them to specialized handlers. 4. Fallback: Handle tasks even if classification fails.
That didn't take long! Nomic AI has finetuned the new ModernBERT-base encoder model into a strong embedding model for search, classification, clustering and more!
Details: 🤖 Based on ModernBERT-base with 149M parameters. 📊 Outperforms both nomic-embed-text-v1 and nomic-embed-text-v1.5 on MTEB! 🏎️ Immediate FA2 and unpacking support for super efficient inference. 🪆 Trained with Matryoshka support, i.e. 2 valid output dimensionalities: 768 and 256. ➡️ Maximum sequence length of 8192 tokens! 2️⃣ Trained in 2 stages: unsupervised contrastive data -> high quality labeled datasets. ➕ Integrated in Sentence Transformers, Transformers, LangChain, LlamaIndex, Haystack, etc. 🏛️ Apache 2.0 licensed: fully commercially permissible
✒️ Ultraset - all-in-one dataset for SFT training in Alpaca format. fluently-sets/ultraset
❓ Ultraset is a comprehensive dataset for training Large Language Models (LLMs) using the SFT (instruction-based Fine-Tuning) method. This dataset consists of over 785 thousand entries in eight languages, including English, Russian, French, Italian, Spanish, German, Chinese, and Korean.
🤯 Ultraset solves the problem faced by users when selecting an appropriate dataset for LLM training. It combines various types of data required to enhance the model's skills in areas such as text writing and editing, mathematics, coding, biology, medicine, finance, and multilingualism.
🤗 For effective use of the dataset, it is recommended to utilize only the "instruction," "input," and "output" columns and train the model for 1-3 epochs. The dataset does not include DPO or Instruct data, making it suitable for training various types of LLM models.
❇️ Ultraset is an excellent tool to improve your language model's skills in diverse knowledge areas.
a new experimental model that unlocks stronger reasoning capabilities and shows its thoughts. The model plans (with thoughts visible), can solve complex problems with Flash speeds, and more
Six predictions for AI in 2025 (and a review of how my 2024 predictions turned out):
- There will be the first major public protest related to AI - A big company will see its market cap divided by two or more because of AI - At least 100,000 personal AI robots will be pre-ordered - China will start to lead the AI race (as a consequence of leading the open-source AI race). - There will be big breakthroughs in AI for biology and chemistry. - We will begin to see the economic and employment growth potential of AI, with 15M AI builders on Hugging Face.
How my predictions for 2024 turned out:
- A hyped AI company will go bankrupt or get acquired for a ridiculously low price ✅ (Inflexion, AdeptAI,...)
- Open-source LLMs will reach the level of the best closed-source LLMs ✅ with QwQ and dozens of others
- Big breakthroughs in AI for video, time-series, biology and chemistry ✅ for video 🔴for time-series, biology and chemistry
- We will talk much more about the cost (monetary and environmental) of AI ✅Monetary 🔴Environmental (😢)
- A popular media will be mostly AI-generated ✅ with NotebookLM by Google
- 10 millions AI builders on Hugging Face leading to no increase of unemployment 🔜currently 7M of AI builders on Hugging Face