Le projet THINK

Intel Labs’ Neural Low-Rank Adapter Search (NLS) produces accurate models with INT4 weights and is available in OpenVINO’s Neural Network Compression Framework

Run Llama 3.3-70B efficiently on Intel® Gaudi® 2 using vLLM. Learn setup, configuration, and performance tips for scalable, production-ready inference.

Learn how to deploy Llama 3.3-70B on Intel® Gaudi® 2 AI accelerators using Hugging Face TGI, with practical setup steps and optimization tips.

Supercharge VLM deployment with TGI on Intel XPUs. This guide shows how to set up, optimize, and serve blazing-fast models using Intel® Tiber AI Cloud.

Let’s take a super fast journey summarizing the strides taken in an era (2012 to 2025 period) from simple image classification to recent video-LLMs to understand how to proceed with “why” questions in video understanding

Unlocking AI’s Potential: Insights from Dell’s Varun Chhabra on Storytelling, Innovation, and Transformation.

Scientists involved in the multi-year research project completed several prototype studies, demonstrating the potential of the program to facilitate student engagement in various learning environments, including classrooms, after-school programs, and home-based learning. The final Kid Space research study shows the value of the immersive learning environment in supporting collaborative problem-solving (CPS) processes in young learners.

Hewlett Packard Enterprises (HPE) set a world record for the TPC Benchmark™ Express AI (TPCx-AI) SF100 benchmark.

Intel Labs offers Atlas CLI, an open source tool for managing machine learning (ML) model provenance, including model artifact integrity and model lineage in ML pipelines

Intel Labs researchers will present eleven papers at conference workshops as part of CVPR 2025. These works include a framework for systematic hierarchical analysis of vision model representations; a flexible graph-learning framework for fine-grained keystep recognition; and a novel interpretability metric that measures how consistently individual attention heads in CLIP models align with specific concepts