Related papers: The AIDA-2020 TLU: a Flexible Trigger Logic Unit f…
Transformers have demonstrated exceptional performance across various domains due to their self-attention mechanism, which captures complex relationships in data. However, training on smaller datasets poses challenges, as standard attention…
The intelligent Distributed Dispatch and Scheduling (iDDS) service is a versatile workflow orchestration system designed for large-scale, distributed scientific computing. iDDS extends traditional workload and data management by integrating…
The risk of soft errors due to radiation continues to be a significant challenge for engineers trying to build systems that can handle harsh environments. Building systems that are Radiation Hardened by Design (RHBD) is the preferred…
Large Language Models (LLMs) exhibit remarkable capabilities in the hierarchical decomposition of complex tasks through semantic reasoning. However, their application in embodied systems faces challenges in ensuring reliable execution of…
Verifying integrity of software execution in low-end micro-controller units (MCUs) is a well-known open problem. The central challenge is how to securely detect software exploits with minimal overhead, since these MCUs are designed for low…
In almost 30 years of operation, the Very Large Array (VLA) has proved to be a remarkably flexible and productive radio telescope. However, the basic capabilities of the VLA have changed little since it was designed. A major expansion…
To enhance the nonlinearity of neural networks and increase their mapping abilities between the inputs and response variables, activation functions play a crucial role to model more complex relationships and patterns in the data. In this…
The ATLAS experiment at the Large Hadron Collider employs a two-level trigger system to record data at an average rate of 1 kHz from physics collisions, starting from an initial bunch crossing rate of 40 MHz. During the LHC Run 2…
This study introduces a novel activation function, characterized by a dynamic slope that adjusts throughout the training process, aimed at enhancing adaptability and performance in deep neural networks for computer vision tasks. The…
We present a highly frequency multiplexed readout for large-format superconducting detector arrays intended for use in the next generation of balloon-borne and space-based sub-millimeter and far-infrared missions. We will demonstrate this…
While Large Language Models (LLMs) have demonstrated remarkable fluency in educational dialogues, most generative tutors primarily operate through intuitive, single-pass generation. This reliance on fast thinking precludes a dedicated…
Thin planar pixel modules are promising candidates to instrument the inner layers of the new ATLAS pixel detector for HL-LHC, thanks to the reduced contribution to the material budget and their high charge collection efficiency after…
Deep learning researchers have a keen interest in proposing two new novel activation functions which can boost network performance. A good choice of activation function can have significant consequences in improving network performance. A…
MALTA is part of the Depleted Monolithic Active Pixel sensors designed in Tower 180nm CMOS imaging technology. A custom telescope with six MALTA planes has been developed for test beam campaigns at SPS, CERN, with the ability to host…
We present Reelay, a unified online temporal logic monitoring framework designed for the rigorous analysis and runtime verification of cyber-physical systems. Reelay addresses the fragmentation of existing logical formalisms and tools by…
SRON (Netherlands Institute for Space Research) is developing the Focal Plane Assembly (FPA) for Athena X-IFU, whose Demonstration Model (DM) will use for the first time a time domain multiplexing (TDM)-based readout system for the on-board…
Since the release of T\"ULU [Wang et al., 2023b], open resources for instruction tuning have developed quickly, from better base models to new finetuning techniques. We test and incorporate a number of these advances into T\"ULU, resulting…
Tool-Integrated Reasoning (TIR) enables large language models (LLMs) to improve their internal reasoning ability by integrating external tools. However, models employing TIR often display suboptimal behaviors, such as insufficient or…
Neurons, modeled as linear threshold unit (LTU), can in theory compute all thresh- old functions. In practice, however, some of these functions require synaptic weights of arbitrary large precision. We show here that dendrites can alleviate…
The dominant paradigm in modern neural networks relies on simple, monotonically-increasing activation functions like ReLU. While effective, this paradigm necessitates large, massively-parameterized models to approximate complex functions.…