Hugging Face is an AI community platform and company providing a hub for open-source machine learning models, datasets, and demo spaces. It hosts over 500,000 models and is widely used by the AI research community.
Hugging Face users are reporting an 'Access Denied' error when attempting to download any model, including open-weight models previously accessible. Multiple users on the Hugging Face forum confirm th…
Robbyant released six embodied-AI models over four days spanning perception, action, world modeling and video, positioning the LingBot stack as a platform for robotics developers. The public artifacts…
MIT researchers have developed a new auditing method called Gaussian probing that can detect whether an AI model has been fine-tuned to generate child sexual abuse material (CSAM) with 100% accuracy w…
A user taking the Hugging Face LLM Course Chapter 11 code quiz reports a DNS resolution error for api-inference.huggingface.co, causing quiz submissions to fail with a MaxRetryError. The user, abhijee…
On May 8, 2026, BerriAI disclosed CVE-2026-42271, a command-injection flaw in LiteLLM, one of the most widely deployed open-source AI gateways. The vulnerability scored CVSS 8.8 and affects releases f…
Hugging Face's parquet API endpoint for datasets has been down for at least five hours, affecting users like a4n9i who rely on it to retrieve parquet file URLs. The issue impacts both the main API and…
Outlines, an open-source library, introduces deterministic certainty into large language model output generation to produce reliable structured outputs such as JSON objects and classification results.…
A developer ported Google's Gemma-4 model family (2B, 4B, 12B) to AWS Inferentia2, achieving up to 44 tok/s on the smallest model. The port required workarounds for three architectural features—mixed …
A roundup of top AI papers on Hugging Face highlights four major themes: real-time interactive video generation with Vidu S1, scientific reasoning on molecular structures via SciReasoner, a critique o…
A developer built a deforestation identification tool using an AI agent, which downloads satellite images, generates NDVI, and identifies potentially deforested locations. The project, created by Shas…
Fine-tuning large language models for social media trends is often the wrong first step, according to a technical analysis that distinguishes between freshness, detection, interpretation, and generati…
A graphic designer with 4,000 social media post layouts is considering organizing them into a structured dataset for training or fine-tuning design models. The dataset, though small for training from …
A hyper-optimized, zero-dependency C/CUDA inference engine for the Qwen 3.6 35B model on RTX 5090 Blackwell GPUs achieves 13.4k tokens/sec prefill throughput at 2,048 context depth and 270+ tokens/sec…
A developer trained a 113M-parameter language model for earthquake science from scratch using open-access papers, Wikipedia, and FineWeb-Edu on two NVIDIA A30 GPUs. The model achieved a 35% reduction …
Open source AI is becoming a cornerstone of innovation for Fortune 500 companies, with Hugging Face emerging as a key platform for collaborative development. CEO Clem Delangue's platform now serves ro…
A team fixing two content tasks that returned off-topic garbage discovered their retrieval-augmented generation (RAG) system was reinforcing its own mistakes. The root cause was an inconsistently appl…
A browser extension can now run a small language model locally, with no server or API key, using either Chrome's built-in Gemini Nano or a custom model via Transformers.js on WebGPU. The approach enab…
Hugging Face's top AI papers for July 12, 2026, highlight trends in video generation, scientific reasoning, video understanding benchmarks, and robotics memory. Vidu S1 enables real-time, voice-contro…
An independent open-source developer has filed an official community complaint against Hugging Face, demanding the reversion of recent free-tier changes that eliminated CPU Basic instances, restricted…
A technical guide advises practitioners to start with supervised fine-tuning (SFT) when correct trajectories are available, treat internal annotation roles like assistant_think as training format rath…