arXiv subjects

Widyadewi Soedarmadji

Publications and source records attributed to Widyadewi Soedarmadji.

4 recordsLinked to original sources

arXiv2026

Virtual Personas for Language Models via an Anthology of Backstories

Large language models (LLMs) are trained from vast repositories of text authored by millions of distinct authors, reflecting an enormous diversity of human traits. While these models bear the potential to be used as approximations of human subjects in behavioral studies, prior efforts have been limited in steering model responses to match individual human users. In this work, we introduce "Anthology", a method for conditioning LLMs to particular virtual personas by harnessing open-ended life narratives, which we refer to as "backstories." We show that our methodology enhances the consistency and reliability of experimental outcomes while ensuring better representation of diverse sub-populations. Across three nationally representative human surveys conducted as part of Pew Research Center's American Trends Panel (ATP), we demonstrate that Anthology achieves up to 18% improvement in matching the response distributions of human respondents and 27% improvement in consistency metrics.

cs.CL↗

arXiv2025

Characterizing and Optimizing Real-Time Optimal Control for Embedded SoCs

Resource-limited robots face significant challenges in executing computationally intensive tasks, such as locomotion and manipulation, particularly for real-time optimal control algorithms like Model Predictive Control (MPC). This paper provides a comprehensive design space exploration to identify optimal hardware computation architectures for these demanding model-based control algorithms. We profile and optimize representative architectural designs, including general-purpose scalar CPUs, vector processors, and specialized accelerators. By characterizing kernel-level benchmarks and end-to-end robotic scenarios, including a hardware-in-the-loop evaluation on a fabricated RISC-V multi-core vector SoC, we present a quantitative comparison of performance, area, and utilization across distinct architectural design points. Our findings demonstrate that targeted architectural modifications, coupled with deep software and system optimizations, enable up to 3.71x speedups for MPC, resulting in up to 27% system-level power reductions while completing robotic tasks. Finally, we propose a code generation flow designed to simplify the complex engineering effort required for mapping robotic workloads onto specialized architectures.

cs.RO↗

arXiv2025

ViDiT-Q: Efficient and Accurate Quantization of Diffusion Transformers for Image and Video Generation

Diffusion transformers have demonstrated remarkable performance in visual generation tasks, such as generating realistic images or videos based on textual instructions. However, larger model sizes and multi-frame processing for video generation lead to increased computational and memory costs, posing challenges for practical deployment on edge devices. Post-Training Quantization (PTQ) is an effective method for reducing memory costs and computational complexity. When quantizing diffusion transformers, we find that existing quantization methods face challenges when applied to text-to-image and video tasks. To address these challenges, we begin by systematically analyzing the source of quantization error and conclude with the unique challenges posed by DiT quantization. Accordingly, we design an improved quantization scheme: ViDiT-Q (Video & Image Diffusion Transformer Quantization), tailored specifically for DiT models. We validate the effectiveness of ViDiT-Q across a variety of text-to-image and video models, achieving W8A8 and W4A8 with negligible degradation in visual quality and metrics. Additionally, we implement efficient GPU kernels to achieve practical 2-2.5x memory saving and a 1.4-1.7x end-to-end latency speedup.

cs.CV↗

arXiv2021

GNSS Radio Occultation on Aerial Platforms with Commercial Off-The-Shelf Receivers

In recent decades, GNSS Radio Occultation soundings have proven an invaluable input to global weather forecasting. The success of government-sponsored programs such as COSMIC is now complemented by commercial low-cost cubesat implementations. The result is access to more than 10,000 soundings per day and improved weather forecasting accuracy. This movement towards commercialization has been supported by several agencies, including the National Aeronautics and Space Administration (NASA), National Oceanic and Atmospheric Administration (NOAA) and the U.S. Air Force (USAF) with programs such as the Commercial Weather Data Pilot (CWDP). This has resulted in further interest in commercially deploying GNSS-RO on complementary platforms. Here, we examine a so far underutilized platform: the high-altitude weather balloon. Such meteorological radiosondes are deployed twice daily at over 900 locations globally and form an essential in-situ data source as a long-standing input to weather forecasting models. Adding GNSS-RO capability to existing radiosonde platforms would greatly expand capability, allowing for persistent and local area monitoring, a feature particularly useful for hurricane and other severe weather monitoring. A prohibitive barrier to entry to this inclusion is cost and complexity as GNSS-RO traditionally requires highly specialized and sensitive equipment. This paper describes a multi-year effort to develop a low-cost and scalable approach to balloon GNSS-RO based on Commercial-Off-The-Shelf (COTS) GNSS receivers. We present hardware prototypes and data processing techniques which demonstrate the technical feasibility of the approach through results from several flight testing campaigns.

eess.SP↗