Large language models (LLMs) have significantly advanced natural language processing, but their progress has yet to be equal across languages. While most LLMs are trained in high-resource languages like English, multilingual models generally underperform monolingual ones. Additionally, aspects of their multilingual foundation sometimes restrict the byproducts they produce, like computational demands and licensing regimes. In this study, we document the development of open-foundation models tailored for use in low-resource settings, their limitations, and their benefits. This is the TeenyTinyLlama pair: two compact models for Brazilian Portuguese text generation. We release them under the permissive Apache 2.0 license on GitHub and Hugging Face for community use and further development.

Fine-tuning large pre-trained language models (LLMs) on particular datasets is a commonly employed strategy in Natural Language Processing (NLP) classification tasks. However, this approach usually results in a loss of models generalizability. In this paper, we present a framework that allows for maintaining generalizability, and enhances the performance on the downstream task by utilizing task-specific context attribution. We show that a linear transformation of the text representation from any transformer model using the task-specific concept operator results in a projection onto the latent concept space, referred to as context attribution in this paper. The specific concept operator is optimized during the supervised learning stage via novel loss functions. The proposed framework demonstrates that context attribution of the text representation for each task objective can improve the capacity of the discriminator function and thus achieve better performance for the classification task. Experimental results on three datasets, namely HateXplain, IMDB reviews, and Social Media Attributions, illustrate that the proposed model attains superior accuracy and generalizability. Specifically, for the non-fine-tuned BERT on the HateXplain dataset, we observe 8% improvement in accuracy and 10% improvement in F1-score. Whereas for the IMDB dataset, fine-tuned state-of-the-art XLNet is outperformed by 1% for both accuracy and F1-score. Furthermore, in an out-of-domain cross-dataset test, DistilBERT fine-tuned on the IMDB dataset in conjunction with the proposed model improves the F1-score on the HateXplain dataset by 7%. For the Social Media Attributions dataset of YouTube comments, we observe 5.2% increase in F1-metric. The proposed framework is implemented with PyTorch and provided open-source on GitHub.

Prompt-based methods have been successfully applied to multilingual pretrained language models for zero-shot cross-lingual understanding. However, most previous studies primarily focused on sentence-level classification tasks, and only a few considered token-level labeling tasks such as Named Entity Recognition (NER) and Part-of-Speech (POS) tagging. In this paper, we propose Token-Level Prompt Decomposition (ToPro), which facilitates the prompt-based method for token-level sequence labeling tasks. The ToPro method decomposes an input sentence into single tokens and applies one prompt template to each token. Our experiments on multilingual NER and POS tagging datasets demonstrate that ToPro-based fine-tuning outperforms Vanilla fine-tuning and Prompt-Tuning in zero-shot cross-lingual transfer, especially for languages that are typologically different from the source language English. Our method also attains state-of-the-art performance when employed with the mT5 model. Besides, our exploratory study in multilingual large language models shows that ToPro performs much better than the current in-context learning method. Overall, the performance improvements show that ToPro could potentially serve as a novel and simple benchmarking method for sequence labeling tasks.

Large Language Models (LLMs) have demonstrated remarkable language understanding and generation capabilities. However, training, deploying, and accessing these models pose notable challenges, including resource-intensive demands, extended training durations, and scalability issues. To address these issues, we introduce a concept of hierarchical, distributed LLM architecture that aims at enhancing the accessibility and deployability of LLMs across heterogeneous computing platforms, including general-purpose computers (e.g., laptops) and IoT-style devices (e.g., embedded systems). By introducing a "layered" approach, the proposed architecture enables on-demand accessibility to LLMs as a customizable service. This approach also ensures optimal trade-offs between the available computational resources and the user's application needs. We envision that the concept of hierarchical LLM will empower extensive, crowd-sourced user bases to harness the capabilities of LLMs, thereby fostering advancements in AI technology in general.

The pervasive spread of misinformation and disinformation poses a significant threat to society. Professional fact-checkers play a key role in addressing this threat, but the vast scale of the problem forces them to prioritize their limited resources. This prioritization may consider a range of factors, such as varying risks of harm posed to specific groups of people. In this work, we investigate potential implications of using a large language model (LLM) to facilitate such prioritization. Because fact-checking impacts a wide range of diverse segments of society, it is important that diverse views are represented in the claim prioritization process. This paper examines whether a LLM can reflect the views of various groups when assessing the harms of misinformation, focusing on gender as a primary variable. We pose two central questions: (1) To what extent do prompts with explicit gender references reflect gender differences in opinion in the United States on topics of social relevance? and (2) To what extent do gender-neutral prompts align with gendered viewpoints on those topics? To analyze these questions, we present the TopicMisinfo dataset, containing 160 fact-checked claims from diverse topics, supplemented by nearly 1600 human annotations with subjective perceptions and annotator demographics. Analyzing responses to gender-specific and neutral prompts, we find that GPT 3.5-Turbo reflects empirically observed gender differences in opinion but amplifies the extent of these differences. These findings illuminate AI's complex role in moderating online communication, with implications for fact-checkers, algorithm designers, and the use of crowd-workers as annotators. We also release the TopicMisinfo dataset to support continuing research in the community.

Training large language models (LLMs) with a large and diverse instruction dataset aligns the models to comprehend and follow human instructions. Recent works have shown that using a small set of high-quality instructions can outperform using large yet more noisy ones. Because instructions are unlabeled and their responses are natural text, traditional active learning schemes with the model's confidence cannot be directly applied to the selection of unlabeled instructions. In this work, we propose a novel method for instruction selection, called SelectLLM, that leverages LLMs for the selection of high-quality instructions. Our high-level idea is to use LLMs to estimate the usefulness and impactfulness of each instruction without the corresponding labels (i.e., responses), via prompting. SelectLLM involves two steps: dividing the unlabelled instructions using a clustering algorithm (e.g., CoreSet) to multiple clusters, and then prompting LLMs to choose high-quality instructions within each cluster. SelectLLM showed comparable or slightly better performance on the popular instruction benchmarks, compared to the recent state-of-the-art selection methods. All code and data are publicly available (https://github.com/minnesotanlp/select-llm).

The intricate relationship between human decision-making and emotions, particularly guilt and regret, has significant implications on behavior and well-being. Yet, these emotions subtle distinctions and interplay are often overlooked in computational models. This paper introduces a dataset tailored to dissect the relationship between guilt and regret and their unique textual markers, filling a notable gap in affective computing research. Our approach treats guilt and regret recognition as a binary classification task and employs three machine learning and six transformer-based deep learning techniques to benchmark the newly created dataset. The study further implements innovative reasoning methods like chain-of-thought and tree-of-thought to assess the models interpretive logic. The results indicate a clear performance edge for transformer-based models, achieving a 90.4% macro F1 score compared to the 85.3% scored by the best machine learning classifier, demonstrating their superior capability in distinguishing complex emotional states.

Text simplification aims to make technical texts more accessible to laypeople but often results in deletion of information and vagueness. This work proposes InfoLossQA, a framework to characterize and recover simplification-induced information loss in form of question-and-answer (QA) pairs. Building on the theory of Question Under Discussion, the QA pairs are designed to help readers deepen their knowledge of a text. We conduct a range of experiments with this framework. First, we collect a dataset of 1,000 linguist-curated QA pairs derived from 104 LLM simplifications of scientific abstracts of medical studies. Our analyses of this data reveal that information loss occurs frequently, and that the QA pairs give a high-level overview of what information was lost. Second, we devise two methods for this task: end-to-end prompting of open-source and commercial language models, and a natural language inference pipeline. With a novel evaluation framework considering the correctness of QA pairs and their linguistic suitability, our expert evaluation reveals that models struggle to reliably identify information loss and applying similar standards as humans at what constitutes information loss.

Peer-to-peer (P2P) lending has emerged as a distinctive financing mechanism, linking borrowers with lenders through online platforms. However, P2P lending faces the challenge of information asymmetry, as lenders often lack sufficient data to assess the creditworthiness of borrowers. This paper proposes a novel approach to address this issue by leveraging the textual descriptions provided by borrowers during the loan application process. Our methodology involves processing these textual descriptions using a Large Language Model (LLM), a powerful tool capable of discerning patterns and semantics within the text. Transfer learning is applied to adapt the LLM to the specific task at hand. Our results derived from the analysis of the Lending Club dataset show that the risk score generated by BERT, a widely used LLM, significantly improves the performance of credit risk classifiers. However, the inherent opacity of LLM-based systems, coupled with uncertainties about potential biases, underscores critical considerations for regulatory frameworks and engenders trust-related concerns among end-users, opening new avenues for future research in the dynamic landscape of P2P lending and artificial intelligence.

An effective multi-turn instruction-following assistant can be developed by creating a simulator that can generate useful interaction data. Apart from relying on its intrinsic weights, an ideal user simulator should also be able to bootstrap external knowledge rapidly in its raw form to simulate the multifarious diversity of text available over the internet. Previous user simulators generally lacked diversity, were mostly closed domain, and necessitated rigid schema making them inefficient to rapidly scale to incorporate external knowledge. In this regard, we introduce, Kaucus, a Knowledge-Augmented User Simulator framework, to outline a process of creating diverse user simulators, that can seamlessly exploit external knowledge as well as benefit downstream assistant model training. Through two GPT-J based simulators viz., a Retrieval Augmented Simulator and a Summary Controlled Simulator we generate diverse simulator-assistant interactions. Through reward and preference model-based evaluations, we find that these interactions serve as useful training data and create more helpful downstream assistants. We also find that incorporating knowledge through retrieval augmentation or summary control helps create better assistants.