Browse every post or search by title and content.
SafeGene introduces reusable safety-adapter modules that restore LLM safety alignment lost during custom fine-tuning, enabling cross-task re...
GitHub published a beginner guide to common platform questions, reflecting a surge in AI developers joining open source. The guide's focus o...
DiBS (Diffusion-Informed Branch Selection) merges diffusion models with symbolic constraint solving to reduce solve times by 73% on Sudoku p...
A small DistilBERT model built during HuggingFace's Build Small hackathon extracts safety warnings from Pakistani government gazettes. It ru...
A new arXiv paper treats AI fairness as a symmetry operation: a classifier is fair if its output is invariant under counterfactual attribute...
HuggingFace’s 'Amazing Digital Dentures' hackathon project failed due to data scarcity and unrealistic AI expectations, offering key lessons...
NVIDIA and Doosan Group expand their partnership to bring physical AI and AI factory infrastructure to robotics, construction, energy, and e...
OpenAI provides free Codex vouchers to participants of Hugging Face's 'Build Small' hackathon, offering developers a risk-free chance to exp...
HuggingFace and five AI labs built a multi-agent finance drama using only small language models under 8B parameters, proving emergent creati...
Researchers introduce Query Retrieve Conclude, a zero-shot framework that interprets evolving internet memes by searching the open web for m...
Vercel now lets developers create Shopify storefronts with one click, using AI tool v0 to generate UI components and auto-configure credenti...
New research on Arxiv shows multi-agent LLM systems waste tokens with unstructured dialogue. A proposed 'action-state' protocol cuts token u...