Search for: All records

Finetuned large language models (LLMs) have shown remarkable performance in financial tasks, such as sentiment analysis and information retrieval. Due to privacy concerns, finetuning and deploying financial LLMs (FinLLMs) locally are crucial for institutions and individuals. In this paper, we employ quantized low-rank adaptation (QLoRA) to finetune FinLLMs, which leverage low-rank structure and quantization technique to significantly reduce computational requirements while maintaining model performance. We also employ data and pipeline parallelism to enable local finetuning on commodity GPUs. Experiments on financial datasets validate the efficacy of our approach in yielding notable improvements over the base models. 

Large Vision-Language Models (LVLMs) have made substantial progress by integrating pre-trained large language models (LLMs) and vision models through instruction tuning. Despite these advancements, LVLMs often exhibit the hallucination phenomenon, where generated text responses appear linguistically plausible but contradict the input image, indicating a misalignment between image and text pairs. This misalignment arises because the model tends to prioritize textual information over visual input, even when both the language model and visual representations are of high quality. Existing methods leverage additional models or human annotations to curate preference data and enhance modality alignment through preference optimization. These approaches are resource-intensive and may not effectively reflect the target LVLM's preferences, making the curated preferences easily distinguishable. Our work addresses these challenges by proposing the Calibrated Self-Rewarding (CSR) approach, which enables the model to self-improve by iteratively generating candidate responses, evaluating the reward for each response, and curating preference data for fine-tuning. In the reward modeling, we employ a step-wise strategy and incorporate visual constraints into the self-rewarding process to place greater emphasis on visual input. Empirical results demonstrate that CSR significantly enhances performance and reduces hallucinations across twelve benchmarks and tasks, achieving substantial improvements over existing methods by 7.62%. Our empirical results are further supported by rigorous theoretical analysis, under mild assumptions, verifying the effectiveness of introducing visual constraints into the self-rewarding paradigm. Additionally, CSR shows compatibility with different vision-language models and the ability to incrementally improve performance through iterative fine-tuning. 

Summary Powdery mildew is an economically important disease caused byc. 1000 different fungal species.Erysiphe vacciniiis an emerging powdery mildew species that is impacting the blueberry industry. Once confined to North America,E. vacciniiis now spreading rapidly across major blueberry‐growing regions, including China, Morocco, Mexico, and the USA, threatening millions in losses.This study documents its recent global spread by analyzing both herbarium specimens, some over 150‐yr‐old, and fresh samples collected world‐wide.Our findings were integrated into a ‘living phylogeny’ via T‐BAS to simplify pathogen identification and enable rapid responses to new outbreaks. We identified 50 haplotypes, two primary introductions world‐wide, and revealed a shift from a generalist to a specialist pathogen.This research provides insights into the complexities of host specialization and highlights the need to address this emerging global threat to blueberry production. 

How do practitioners who develop consumer AI products scope, motivate, and conduct privacy work? Respecting pri- vacy is a key principle for developing ethical, human-centered AI systems, but we cannot hope to better support practitioners without answers to that question. We interviewed 35 industry AI practitioners to bridge that gap. We found that practitioners viewed privacy as actions taken against pre-defined intrusions that can be exacerbated by the capabilities and requirements of AI, but few were aware of AI-specific privacy intrusions documented in prior literature. We found that their privacy work was rigidly defined and situated, guided by compliance with privacy regulations and policies, and generally demoti- vated beyond meeting minimum requirements. Finally, we found that the methods, tools, and resources they used in their privacy work generally did not help address the unique pri- vacy risks introduced or exacerbated by their use of AI in their products. Collectively, these findings reveal the need and opportunity to create tools, resources, and support structures to improve practitioners’ awareness of AI-specific privacy risks, motivations to do AI privacy work, and ability to ad- dress privacy harms introduced or exacerbated by their use of AI in consumer products. 

Predicting the behavior of nanomaterials under various conditions presents a significant challenge due to their complex microstructures. While high-fidelity modeling techniques, such as molecular dynamics (MD) simulations, are effective, they are also computationally demanding. Machine learning (ML) models have opened new avenues for the rapid exploration of design spaces. In this work, we developed a deep learning framework based on a conditional generative adversarial network (cGAN) to predict the evolution of grain boundary (GB) networks in nanocrystalline materials under mechanical loads, incorporating both morphological and atomic details. We conducted MD simulations on nanocrystalline tungsten and used the resulting ground-truth data to train our cGAN model. We assessed the performance of our cGAN model by comparing it to a Convolutional Autoencoder (ConvAE) model and examining the impact of changes in geometric morphology and loading conditions on the model's performance. Our cGAN model demonstrated high accuracy in predicting GB network evolution under a variety of conditions. This developed framework shows potential for predicting various materials' behaviors across a wide range of nanomaterials. 

Abstract Nickel stable isotopes (δ⁶⁰Ni) provide insight to Ni biogeochemistry in the modern and past oceans. Here, we present the first Pacific Ocean high‐resolution dissolved Ni concentration and δ⁶⁰Ni data, from the US GEOTRACES GP15 cruise. As in other ocean basins, increases in δ⁶⁰Ni toward the surface ocean are observed across the entire transect, reflecting preferential biological uptake of light Ni isotopes, however the observed magnitude of fractionation is larger in the tropical Pacific than the North Pacific Subtropical Gyre. Such surface ocean fractionation by phytoplankton should accumulate isotopically lighter Ni in the deep Pacific, yet we find that North Pacific deep ocean δ⁶⁰Ni is similar to previously reported values from the deep Atlantic. Finally, we find that seawater dissolved δ⁶⁰Ni in regions with hydrothermal input can be either higher or lower than background deep ocean δ⁶⁰Ni, depending on vent geochemistry and proximity.