Search for: All records

Abstract As innovation in deep learning continues, many engineers are incorporating Pre-Trained Models (PTMs) as components in computer systems. Some PTMs are foundation models, and others are fine-tuned variations adapted to different needs. When these PTMs are named well, it facilitates model discovery and reuse. However, prior research has shown that model names are not always well chosen and can sometimes be inaccurate and misleading. The naming practices for PTM packages have not been systematically studied, which hampers engineers’ ability to efficiently search for and reliably reuse these models. In this paper, we conduct the first empirical investigation of PTM naming practices in the Hugging Face PTM registry. We begin by reporting on a survey of 108 Hugging Face users, highlighting differences from traditional software package naming and presenting findings on PTM naming practices. The survey results indicate a mismatch between engineers’ preferences and current practices in PTM naming. We then introduce DARA, the first automatedDNNARchitectureAssessment technique designed to detect PTM naming inconsistencies. Our results demonstrate that architectural information alone is sufficient to detect these inconsistencies, achieving an accuracy of 94% in identifying model types and promising performance (over 70%) in other architectural metadata as well. We also highlight potential use cases for automated naming tools, such as model validation, PTM metadata generation and verification, and plagiarism detection. Our study provides a foundation for automating naming inconsistency detection. Finally, we envision future work focusing on automated tools for standardizing package naming, improving model selection and reuse, and strengthening the security of the PTM supply chain.“The main idea is to treat a program as a piece of literature, addressed to human beings rather than to a computer”—D. Knuth 

The alignment of large language models (LLMs) with human values is critical as these models become increasingly integrated into various societal and decision-making processes. Traditional methods, such as reinforcement learning from human feedback (RLHF), achieve alignment by fine-tuning model parameters, but these approaches are often computationally expensive and impractical when models are frozen or inaccessible for parameter modification. In contrast, prompt optimization is a viable alternative to RLHF for LLM alignment. While the existing literature has shown empirical promise of prompt optimization, its theoretical underpinning remains under-explored. We address this gap by formulating prompt optimization as an optimization problem and try to provide theoretical insights into the optimality of such a framework. To analyze the performance of the prompt optimization, we study theoretical suboptimality bounds and provide insights in terms of how prompt optimization depends upon the given prompter and target model. We also provide empirical validation through experiments on various datasets, demonstrating that prompt optimization can effectively align LLMs, even when parameter fine-tuning is not feasible. 

We propose Vision Token Turing Machines (ViTTM), an efficient, low-latency, memory-augmented Vision Transformer (ViT). Our approach builds on Neural Turing Machines and Token Turing Machines, which were applied to NLP and sequential visual understanding tasks. ViTTMs are designed for non-sequential computer vision tasks such as image classification and segmentation. Our model creates two sets of tokens: process tokens and memory tokens; process tokens pass through encoder blocks and read-write from memory tokens at each encoder block in the network, allowing them to store and retrieve information from memory. By ensuring that there are fewer process tokens than memory tokens, we are able to reduce the inference time of the network while maintaining its accuracy. On ImageNet-1K, the state-of-the-art ViT-B has median latency of 529.5ms and 81.0% accuracy, while our ViTTM-B is 56% faster (234.1ms), with 2.4 times fewer FLOPs, with an accuracy of 82.9%. On ADE20K semantic segmentation, ViT-B achieves 45.65mIoU at 13.8 frame-per-second (FPS) whereas our ViTTM-B model acheives a 45.17 mIoU with 26.8 FPS (+94%). 

Discrete depth profiles of water temperature, dissolved oxygen, oxidation-reduction potential, conductivity, specific conductance, and pH were collected with multiple handheld water quality probes and discrete depth profiles of photosynthetically active radiation (PAR) were collected with a LI-COR underwater light meter from 2013 to 2024 in five drinking water reservoirs in southwestern Virginia, USA. These reservoirs are: Beaverdam Reservoir (Vinton, Virginia), Carvins Cove Reservoir (Roanoke, Virginia), Falling Creek Reservoir (Vinton, Virginia), Gatewood Reservoir (Pulaski, Virginia), and Spring Hollow Reservoir (Salem, Virginia). Beaverdam, Carvins Cove, Falling Creek, and Spring Hollow Reservoirs are owned and operated by the Western Virginia Water Authority as primary or secondary drinking water sources for Roanoke, Virginia, and Gatewood Reservoir is a drinking water source for the Town of Pulaski, Virginia. All discrete depth profiles were collected on approximately 1-meter intervals. The data package consists of two datasets: 1) Secchi depth data; and 2) discrete depth profiles of multiple water quality variables measured by handheld sensors. The Secchi data and discrete depth profiles were measured at the deepest site of each reservoir adjacent to the dam, as well as other in-reservoir sites. Handheld sensor measurements were also collected at a gauged weir on the primary inflow tributary, other inflows, and outflows at Falling Creek Reservoir; inflows and outflows at Beaverdam Reservoir; and inflows at Carvins Cove Reservoir. In 2021, YSI handheld data were also collected from a littoral site in Beaverdam Reservoir. Data were collected approximately fortnightly in the spring months (March - May), weekly in the summer and early autumn (June - September), and monthly in the late autumn and winter (October - February) in Falling Creek and Beaverdam Reservoirs; data coverage in the other three reservoirs varies among years. Note there are some YSI depth profiles and Secchi observations that were measured at night during overnight sampling. All of these observations have the correct time associated with them. There was a major revision of this dataset from its previous version, which included correcting times of observations, including values of negative ORP, and adding observations from an outflow at Falling Creek Reservoir.