We are pleased to announce that Embedl will be exhibiting at the EurIPS 2025 Startup Village, taking place in Copenhagen from December 3 to 5. EurIPS is a meeting point for researchers, engineers, and companies working in machine learning and artificial intelligence. Participating in the Startup Village provides an opportunity to present our work, exchange insights, and engage with the wider research and innovation community.
At Embedl, we focus on making deep learning models more efficient to train, run, and deploy. As models grow in size and computational demands increase, systematic optimization becomes crucial for achieving optimal performance, minimizing energy consumption, and ensuring scalability across various hardware platforms.
During the event, we will demonstrate:
Model compression techniques, including quantization and pruning
Hardware-aware optimizations that enable efficient deployment on diverse platforms
Tools for reproducible benchmarking and deployment evaluation
Our aim is to highlight methods that are both scientifically grounded and practically applicable in production environments.
Alongside our established work on optimization for CNNs and transformer-based models, we will also present an update on our current developments in large language model (LLM) optimization.
This work focuses on:
Improving inference efficiency for LLM architectures
Reducing memory consumption while maintaining model quality
Ensuring stable and predictable performance across multiple hardware backends
These developments build on our long-standing efforts in scalable model optimization and address the growing need for reliable and cost-efficient LLM deployment.
We welcome anyone interested in model efficiency, deployment workflows, or the evolving landscape of AI optimization to visit our booth. Whether you're exploring ways to reduce computational cost, evaluating optimization strategies, or simply learning about current approaches, we look forward to discussing our work and hearing about your challenges and perspectives.
We look forward to meeting the community at EurIPS and contributing to ongoing discussions about efficient and scalable AI. Visit our booth or schedule a meeting HERE.