We’re on a journey to advance and democratize artificial intelligence through open source and open science.

HuggingFace's platform is a resource for developers and researchers working in natural language processing (NLP) and machine learning, offering insights into NLP models, tools, and datasets. Through articles, tutorials, and open-source projects, HuggingFace offers insights into state-of-the-art NLP techniques, transformer architectures, and transfer learning methods. Developers can learn about using pre-trained models, fine-tuning strategies, and deploying NLP applications with HuggingFace's libraries and APIs.

Hugging Face

SmolVLM has introduced two new models, SmolVLM-256M and SmolVLM-500M, which are designed to be extremely efficient while maintaining strong multimodal performance. These models are significantly smaller than their predecessors, with 256M and 500M parameters, respectively. They can be used with transformers, MLX, and ONNX, and are optimized for tasks such as captioning, document Q&A, and visual reasoning. The release includes four checkpoints and offers greater efficiency, making these models ideal for use on constrained devices or in cost-effective data processing.

SmolVLM Grows Smaller – Introducing the 256M & 500M Models!

TLDR Table of Contents Overview Why Go Smaller? Meet the 256M Parameter Giant A Step Up: 500M What Changed Since SmolVLM 2B? Smaller Multimodal Retrieval: ColSmolVLM 256M & 500M SmolDocling Using Smaller SmolVLM Next Steps TLDR

Smaller Multimodal Retrieval: ColSmolVLM 256M & 500M