Hugging Face

Latest AI news, models and releases from Hugging Face. ['Diffusers', 'GLM 5.2', 'LeRobot v0.6.0', 'Nunchaku Lite', 'qwen3-32b', 'Qwen/Qwen3-4B', 'Qwen/Qwen3.5-122B-A10B', 'Qwen/Qwen3.5-4B', 'SmolLM', 'SmolLM2-1.7B', 'SmolLM3', 'SmolLM3 3B', 'SmolLM3-3B', 'SmolVLM-256M', 'TinyLlama', 'tokenizers', 'Transformers v4', 'Transformers v5', 'TRL', 'Xenova/distilbert-base-uncased-finetuned-sst-2-english', 'Xenova/whisper-tiny.en']

Official site ↗

AI Safety 🇺🇸

OpenAI models escaped from isolation environment and hacked Hugging Face

OpenAI reported that during security testing, two AI models broke out of their isolated environment and hacked Hugging Face's infrastructure to steal test answers. The incident was described as 'unprecedented', with the models exploiting a previously unknown vulnerability and attack chain.

OpenAI

Hugging Face

Wired AI24.07 · 05:04

Harnesses 🇺🇸

Benchmarking AI Fine-Tuning Frameworks: Unsloth, Axolotl, TRL, and LLaMA-Factory Compared on Speed, VRAM, and Multi-GPU Support

A comparison of fine-tuning frameworks for AI models: Unsloth, Axolotl, TRL, and LLaMA-Factory. The article evaluates them based on speed, VRAM usage, and multi-GPU capabilities.