AI
GPT-4o mini: advancing cost-efficient intelligence
OpenAI has released a new version of its GPT-4 model, called GPT-4o mini. This updated model is designed to be more cost-efficient while maintaining t...
AI
AI
New compliance and administrative tools for ChatGPT Enterprise
OpenAI has released new tools for its ChatGPT Enterprise platform. These include a Compliance API that integrates with existing systems, SCIM (System ...
AI
Docmatix - a huge dataset for Document Visual Question Answering
Docmatix is a large-scale dataset designed to facilitate research in Document Visual Question Answering (VQA). It contains over 150,000 images of docu...
AI
TGI Multi-LoRA: Deploy Once, Serve 30 Models
Researchers have developed a technique called TGI Multi-LoRA that allows for the deployment of a single model to serve multiple tasks. This is achieve...
AI
Prover-Verifier Games improve legibility of language model outputs
Prover-verifier games are a new technique used in natural language processing that improves the clarity and trustworthiness of language model outputs....
AI
SmolLM - blazingly fast and remarkably powerful
SmolLM is a new language model that has shown impressive performance on various tasks, including question-answering and text generation. It achieves t...
AI
How we leveraged distilabel to create an Argilla 2.0 Chatbot
Researchers created a chatbot using the Hugging Face library's DistillBERT model, called Argilla 2.0. They used this pre-trained language model to imp...
AI
How NuminaMath Won the 1st AIMO Progress Prize
NuminaMath won the first AIMO Progress Prize, a competition that recognizes advancements in artificial general intelligence. The prize was awarded to ...
AI
OpenAI and Los Alamos National Laboratory announce research partnership
OpenAI has partnered with the Los Alamos National Laboratory to develop safety evaluations for advanced AI models. The goal is to assess and measure p...
AI
Announcing New Hugging Face and KerasHub integration
Hugging Face, a popular platform for natural language processing (NLP) models, has announced the integration of its model hub with KerasHub. This inte...
AI
Experimenting with Automatic PII Detection on the Hub using Presidio
Researchers at Hugging Face are experimenting with automatic detection of personally identifiable information (PII) on their platform, the Hub. They'r...
AI
Preference Optimization for Vision Language Models
Researchers have proposed a method called Differential Private Optimization (DPO) to improve the performance of vision language models. DPO is designe...