Loading…
Summer at Hugging Face
system
Summary
Hugging Face released several major updates across its ecosystem, expanding its public Hub repository to over 16,000 models. Platform additions include Spaces for deploying Gradio and Streamlit demo applications, automatic TensorBoard instances, and evaluation metric tracking integrated with Papers With Code leaderboards. The Transformers library gained JAX/Flax support across more than 5,000 models, improved TensorFlow implementations, and introduced the transformers.onnx export module for model conversion. In research, the BigScience project completed large-scale training of a 13-billion-parameter English decoder model on Jean Zay, while the DeDLOC method enabled training the sahajBERT Bengali model without HPC infrastructure. Accepted conference papers detailed the Datasets library, prompt data-point equivalence, prompt-based fine-tuning heuristics, and block pruning techniques that reduced BERT size by 74 percent.
Context
Hugging Face aimed to empower the machine learning community through open and collaborative technology by expanding platform capabilities, open-source tooling, educational resources, and research programs.
Approach / What changed
The team introduced Spaces for hosting Python ML demos, integrated TensorBoard and Papers With Code metric tracking on the Hub, added JAX/Flax and ONNX export support to Transformers, organized community training sprints, and published research on distributed training, prompting, and block pruning.
Takeaways
- Hugging Face Spaces provides free hosting for machine learning demo applications using Python SDKs like Gradio and Streamlit, supporting secrets, custom requirements, and GitHub repository synchronization.
- The Transformers library added JAX/Flax as its third supported framework with over 5,000 models, upgraded its TensorFlow examples, and introduced the transformers.onnx module for direct model export.
- The DeDLOC research collaboration created a distributed method to train large neural networks collaboratively across volunteer machines, Colaboratory, and preemptible VMs without HPC clusters.
Related reading
Transformers v5: Simple model definitions powering the AI ecosystem
Transformers v5.0.0rc-0 introduces major architectural updates focused on simplicity, training, inference, and ecosystem interoperability across modern AI workflows. The release adopts a modular modeling approach and centralizes attention implementations into a unified AttentionInterface abstraction to reduce contribution and code review overhead. Support for Flax and TensorFlow is officially sunset in favor of focusing on PyTorch as the primary backend, while tokenization is standardized around the tokenizers library. For execution workloads, v5 adds native continuous batching, paged attention mechanisms, and a dedicated transformers serve OpenAI-compatible serving system. Finally, weight loading is refactored to make low-precision quantization a first-class citizen alongside broad interoperability with formats such as GGUF, MLX, and TorchAO.
Lysandre, Arthur Zucker, Cyril Vallez, Vaibhav Srivastav