Mixture-of-Experts (MoE) models scale efficiently by activating only a subset of experts per token, offering a computationally sparse alternative to dense architectures. While prior post-training optimizations, such as inter- and intra-expert pruning, reduce memory usage they provide limited gains in inference-time compute efficiency. Moreover, existing MoE architectures typically activate a fixed number of experts uniformly across all layers, resulting in redundant computation and suboptimal performance. In this work, we first demonstrate that MoE pruning strategies improve only the memory footprint but do not significantly improve inference performance on GPU using optimized frameworks such as vLLM. To address this, we introduce LExI, a data-free optimization technique that determines the optimal number of active experts per layer in a pretrained MoE model. LExI leverages only the model weights to estimate the relative importance of each layer and adaptively assigns the number of active experts accordingly per layer. Experiments on state-of-the-art language and vision MoE benchmarks demonstrate that LExI significantly outperforms traditional MoE pruning approaches in terms of inference efficiency with negligible accuracy loss. For example, using LExI, Qwen1.5-MoE achieves the same throughput on Nvidia H100 GPU with 10% better accuracy than traditional expert pruning.

Using the full potential of LLMs in higher education is hindered by challenges with access to LLMs. The two main access modes currently discussed are paying for a cloud-based LLM or providing a locally maintained open LLM. In this paper, we describe the current state of establishing an open LLM infrastructure at FernUniversit\"at in Hagen under the project name FLEXI (FernUni LLM Experimental Infrastructure). FLEXI enables experimentation within teaching and research with the goal of generating strongly needed evidence in favor (or against) the use of locally maintained open LLMs in higher education. The paper will provide some practical guidance for everyone trying to decide whether to run their own LLM server.

WEHEAD connects to ChatGPT and displays a face, expressions and voice. The voice Alexis "Lexi" Bogan had before last summer was exuberant. She loved to belt out Taylor Swift and Zach Bryan ballads in the car. She laughed all the time -- even while corralling misbehaving preschoolers or debating politics with friends over a backyard fire pit. In high school, she was a soprano in the chorus.

Humans can learn to operate the user interface (UI) of an application by reading an instruction manual or how-to guide. Along with text, these resources include visual content such as UI screenshots and images of application icons referenced in the text. We explore how to leverage this data to learn generic visio-linguistic representations of UI screens and their components. These representations are useful in many real applications, such as accessibility, voice navigation, and task automation. Prior UI representation models rely on UI metadata (UI trees and accessibility labels), which is often missing, incompletely defined, or not accessible. We avoid such a dependency, and propose Lexi, a pre-trained vision and language model designed to handle the unique features of UI screens, including their text richness and context sensitivity. To train Lexi we curate the UICaption dataset consisting of 114k UI images paired with descriptions of their functionality. We evaluate Lexi on four tasks: UI action entailment, instruction-based UI image retrieval, grounding referring expressions, and UI entity recognition.

Opened in November 2020, Hotel Sky in Sandton, Johannesburg, made its debut with three robots: Lexi, Micah, and Ariel. Lending a helpful hand to the human staff at the property, these robots are the hotel's answer to travelers' increased desire for socially distant interactions. Lexi, Micah, and Ariel can deliver room service, provide travel information, and carry up to 165 pounds of luggage each from the marble-floored lobby to the rooms.

Large ontologies still pose serious challenges to state-of-the-art ontology alignment systems. In this paper we present an approach that combines a neural embedding model and logic-based modules to accurately divide an input ontology matching task into smaller and more tractable matching (sub)tasks. We have conducted a comprehensive evaluation using the datasets of the Ontology Alignment Evaluation Initiative. The results are encouraging and suggest that the proposed method is adequate in practice and can be integrated within the workflow of systems unable to cope with very large ontologies.

The system is the first of its kind worldwide and has the potential to help roughly 400,000 Danes with dyslexia and other reading difficulties. Personalized text simplification - that's the gist of a recently completed PhD project from the University of Copenhagen's Department of Computer Science. Postdoc Joachim Bingel's system makes it easier for dyslexics and others with reading difficulties to read texts online. The artificial intelligence-based software replaces difficult words in sentences with simpler alternatives, while learning which words, endings, etc. a user is having particular difficulty with. "We live in a knowledge-based society where anyone without access to knowledge and information due to reading or language difficulties is quickly sidelined," according to head researcher, Postdoc Joachim Bingel.

APAC regional legal technology conference LexTech has launched an AI assistant to help facilitate a better conference experience, making its services available through Facebook Messenger -- a first in the local legal tech industry. Called, LEXi, the Artificial Intelligence chatbot was developed by Malaysia's CanChat and will focus on addressing enquiries of LexTech delegates in real-time via Facebook Messenger. Notably, between now and August 31, 2018, potential conference delegates can secure a 20% discount code by asking LEXi for one. Lee Su Wen, Director of the LexTech Conference 2018 said that: "By leveraging on AI capabilities, the chatbot is a smart and intuitive way to provide information and directions to our audience with great convenience and speed. With LEXi taking on these tasks, we are now able to more efficiently re-allocate our manpower to the more strategically-intensive departments, and by extension creating a better experience for all delegates."

If you've debating building a DIY prototype of the Echo so you don't have to pay a hefty 180 for the convenience to use Amazon's voice assistant Alexa, Lexi might just be the app you've been looking for. Available for iOS, Lexi lets you speak to Alexa straight through your phone – and without the need for any third-party gadgets like the Echo or Triby. Our biggest ever edition of TNW Conference is fast approaching! What's particularly nifty about Lexi is that the app supports a wide range of the Echo's functionalities like placing orders, remotely controlling your smart home devices as well as install and turn skills from the Alexa App. Additionally, Lexi also allows users to ask Alexa for information about the the latest news, weather, movies or directions to the closest hospital, for instance.

As someone who recently bought--and was quickly enchanted by--Amazon's Echo connected speaker, I've been thinking of ways to get its Alexa virtual assistant into more rooms of the house. Of course, I could just buy another Echo speaker, or some other Alexa-infused device such as the Echo Dot, Amazon Tap, or Invoxia's Triby. But what if I could just use my smartphone to access Alexa instead? Or better yet, what if I could repurpose an old smartphone to act like an Echo full-time? Such is the promise of a couple new smartphone apps that have Amazon's Alexa assistant built-in.