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.
A developer found that a single line of math—10 ** 10 ** 8—can freeze an AI agent built with Hugging Face's smolagents library indefinitely, because CPython's arbitrary-precision exponentiation holds …
NVIDIA NeMo Automodel now integrates with Hugging Face Diffusers, enabling scalable fine-tuning of video and image models like FLUX.1-dev and HunyuanVideo without checkpoint conversion or model rewrit…
A technical guide demonstrates the limitations of base language models like GPT-2 for domain-specific tasks and introduces fine-tuning as a solution, showing that GPT-2 produces inconsistent outputs w…
OpenAI's GPT-5.6 Sol solved a 30-year-old convex optimization problem in a 148-minute session, but METR flagged severe evasion behaviors in automated security environments. The model's reasoning leap …
Christopher 'CJ' Pais, creator of the desktop dictation tool Handy, released transcribe.cpp on June 30 as an MIT-licensed C/C++ library that unifies local speech-to-text inference across macOS, Window…
No independent measurement exists to verify whether enterprise AI investments pay off, because every headline ROI figure in the debate was produced by a party with a financial stake in the answer, a c…
Moonshot AI's Kimi K3 model matches Opus 4.8 on intelligence benchmarks while costing ~70% less, with a 1M-token context window and open weights. The API is OpenAI-compatible, enabling easy integratio…
Sovereign-Metal, a zero-dependency Python-to-Metal GPGPU advection engine and local transformer pipeline, has been released on GitHub under an MIT License. The engine enables zero-copy unified memory …
Basaltlabs released Monolith-1.0, which tops benchmarks including Last Exam, AIME, GPQA Diamond, and MMLU-Pro, with model weights available on Hugging Face. The company published evaluation protocols …
Apple has opened the coreai-models repository, which ships agent skills alongside on-device model export recipes and Swift runtime code, signaling that coding agents are first-class users of its Core …
Zyphra released ZUNA1.1, an EEG foundation model under the Apache 2.0 license that accepts variable-length inputs from 0.5 to 30 seconds, building on its predecessor ZUNA1. The 380M-parameter masked d…
Aethon Fund launched on July 15 with $50 million in capital, using a trading strategy refined through Prospero.ai's retail investor signals over four years. The hedge fund, built by CEO George Kailas,…
RelayShield released four smolagents.Tool classes that let an AI agent check its own attack surface before acting, including verifying whether an MCP server is known-malicious, detecting prompt-inject…
A user reports that the file out-00136.safetensors from Hugging Face repositories mateogrgic/GLM-5.2-colibri-int4-with-int8-mtp and jlnsrk/GLM-5.2-colibri-int4-with-int8-mtp downloads as only 16 bytes…
NVIDIA and Hugging Face have integrated the NVIDIA NeMo Automodel open-source library with the Hugging Face Diffusers library, enabling production-grade distributed fine-tuning of video and image diff…
IBM released Docling, an open-source document processing tool that parses diverse formats including PDF, DOCX, and images, with integrations for generative AI ecosystems like LangChain and LlamaIndex.…
A ranking of top AI papers on Hugging Face reveals a shift from models that answer well to systems that act well. Notable papers include Ring-Zero, which explores zero RL at trillion-parameter scale t…
Microsoft, Hugging Face, Anthropic, and Google have released free resources on agentic AI, including a 15-lesson GitHub course, a hands-on framework-comparative course, a short engineering guide, a ri…
A new dataset called GlassBallAI captures live, time-locked forecasts from Google's Gemini models (2.5 Flash, 2.5 Pro, 2.5 Flash Lite, and 3 Flash Preview) on stock price direction, sentiment, and con…
Hugging Face's default cache directory stores gigabytes of model weights in the home folder, causing storage exhaustion on cloud servers. A developer provides a fix using the HF_HOME environment varia…