26 articles
OpenAI shares lessons from a limited preview of Voice Engine, its model for creating custom synthetic voices. The article discusses ethical challenges and opportunities of technology that can generate voices from short audio samples.
Zelma uses GPT-4 to make education data accessible. The tool helps structure and leverage educational data through generative AI.
OpenAI shares early feedback from artists collaborating with Sora to understand how the video generation model integrates into creative workflows.
Hugging Face introduces Pollen-Vision, a unified interface for zero-shot vision models in robotics. The tool integrates multiple vision models (CLIP, DINOv2, SAM) into a common framework, enabling robots to perform perception tasks without task-specific training.
Hugging Face introduces binary and scalar embedding quantization to accelerate and reduce costs for vector retrieval. The method compresses dense representations while maintaining information retrieval quality.
Beginner's guide to Hugging Face Transformers. Covers fundamental concepts, library usage, and common use cases without requiring advanced technical background.
JetBrains integrates OpenAI's API into its developer tools to build its fastest-growing product ever. AI integration accelerates developer workflows through code generation and analysis capabilities.
Hugging Face introduces Chatbot Guardrails Arena, a benchmarking platform to evaluate the robustness of chatbot safety guardrails against adversarial attacks and jailbreak attempts.
GaLore reduces GPU memory requirements for large model training through gradient projection decomposition. Enables training of 7B-70B models on consumer hardware (RTX 4090) with speedups up to 65% versus standard methods.
Hugging Face demonstrates running Phi-2 on Intel Meteor Lake CPU without GPU. The model runs locally on laptop with acceptable performance for conversational chatbot use.
Hugging Face introduces Cosmopedia, a method for generating large-scale synthetic data for LLM pre-training. The dataset contains 30 billion tokens generated via Mixtral 8x7B, covering mathematics, science, and programming. Models trained on this data achieve performance comparable to models pre-trained on natural data.
Superhuman partners with OpenAI to reimagine email experience using AI. The platform integrates OpenAI's capabilities to enhance email management and composition.
Salesforce integrates OpenAI's enterprise-ready LLMs into customer applications. This integration aims to transform customer relationship management capabilities with secure and reliable models for enterprise environments.
Holiday Extras rolls out ChatGPT Enterprise across all departments, gaining 500 hours of productivity weekly. The company adopts a data-driven approach to measure AI's impact on operations.
Hugging Face releases Quanto, a PyTorch quantization backend integrated into Optimum. The tool reduces model size and accelerates inference through quantization, compatible with popular transformer models.
Hugging Face integrates NVIDIA DGX Cloud to simplify model training on H100 GPUs. Users access the infrastructure directly through Hugging Face platform without complex setup.
Hugging Face and Intel optimize embeddings for CPU using Optimum Intel and fastRAG. The solution reduces latency and memory consumption of embedding models on processors, improving RAG performance without GPU.
Hugging Face introduces WebSight, a dataset for converting web screenshots into HTML code. The initiative aims to train vision models capable of generating code from user interface images.
OpenAI partners with Le Monde and Prisa Media to integrate French and Spanish news content into ChatGPT.
Healthify partners with OpenAI to develop AI health coaching aimed at improving sustainable weight loss for millions of people.
OpenAI announces three new board members: Sue Desmond-Hellmann, Nicole Seligman, and Fidji Simo. Sam Altman rejoins the board.
OpenAI completes internal review. Sam Altman and Greg Brockman to continue leading the company. New board members appointed and governance structure enhancements announced.
Paradigm uses OpenAI's API to improve patient access to clinical trials. The solution helps identify eligible patients and connect them to appropriate studies.
Lifespan uses GPT-4 to significantly improve health literacy and patient outcomes.
Hugging Face introduces ConTextual, a benchmark to evaluate how well multimodal models jointly reason over text and images in text-rich scenes. It measures fine-grained understanding of models when handling text embedded within images.
Hugging Face and Argilla partner to enable communities to collectively build higher-quality datasets through Hugging Face Spaces. The platform streamlines collaborative annotation and data sharing to improve AI models.