Semi-structured explanation depicts the implicit process of a reasoner with an explicit representation. This explanation highlights how available information in a specific query is utilised and supplemented with information a reasoner produces from its internal weights towards generating an answer. Despite the recent improvements in generative capabilities of language models, producing structured explanations to verify a model's true reasoning capabilities remains a challenge. This issue is particularly pronounced for not-so-large LMs (e.g., FLAN-T5-XXL). In this work, we first underscore the limitations of supervised fine-tuning (SFT) in tackling this challenge, and then introduce a carefully crafted reward engineering method in reinforcement learning (RL) to better address this problem. We investigate multiple reward aggregation methods and provide a detailed discussion which sheds light on the promising potential of RL for future research. Our proposed method on two semi-structured explanation generation benchmarks (ExplaGraph and COPA-SSE) achieves new state-of-the-art results.

Background Large Language Models (LLMs), enhanced with Clinical Practice Guidelines (CPGs), can significantly improve Clinical Decision Support (CDS). However, methods for incorporating CPGs into LLMs are not well studied. Methods We develop three distinct methods for incorporating CPGs into LLMs: Binary Decision Tree (BDT), Program-Aided Graph Construction (PAGC), and Chain-of-Thought-Few-Shot Prompting (CoT-FSP). To evaluate the effectiveness of the proposed methods, we create a set of synthetic patient descriptions and conduct both automatic and human evaluation of the responses generated by four LLMs: GPT-4, GPT-3.5 Turbo, LLaMA, and PaLM 2. Zero-Shot Prompting (ZSP) was used as the baseline method. We focus on CDS for COVID-19 outpatient treatment as the case study. Results All four LLMs exhibit improved performance when enhanced with CPGs compared to the baseline ZSP. BDT outperformed both CoT-FSP and PAGC in automatic evaluation. All of the proposed methods demonstrated high performance in human evaluation. Conclusion LLMs enhanced with CPGs demonstrate superior performance, as compared to plain LLMs with ZSP, in providing accurate recommendations for COVID-19 outpatient treatment, which also highlights the potential for broader applications beyond the case study.

This paper investigates the ability of transformer-based models to learn structural recursion from examples. Recursion is a universal concept in both natural and formal languages. Structural recursion is central to the programming language and formal mathematics tasks where symbolic tools currently excel beyond neural models, such as inferring semantic relations between datatypes and emulating program behavior. We introduce a general framework that nicely connects the abstract concepts of structural recursion in the programming language domain to concrete sequence modeling problems and learned models' behavior. The framework includes a representation that captures the general \textit{syntax} of structural recursion, coupled with two different frameworks for understanding their \textit{semantics} -- one that is more natural from a programming languages perspective and one that helps bridge that perspective with a mechanistic understanding of the underlying transformer architecture. With our framework as a powerful conceptual tool, we identify different issues under various set-ups. The models trained to emulate recursive computations cannot fully capture the recursion yet instead fit short-cut algorithms and thus cannot solve certain edge cases that are under-represented in the training distribution. In addition, it is difficult for state-of-the-art large language models (LLMs) to mine recursive rules from in-context demonstrations. Meanwhile, these LLMs fail in interesting ways when emulating reduction (step-wise computation) of the recursive function.

Aligning large language models with human objectives is paramount, yet common approaches including RLHF suffer from unstable and resource-intensive training. In response to this challenge, we introduce ARGS, Alignment as Reward-Guided Search, a novel framework that integrates alignment into the decoding process, eliminating the need for expensive RL training. By adjusting the model's probabilistic predictions using a reward signal, ARGS generates texts with semantic diversity while being aligned with human preferences, offering a promising and flexible solution for aligning language models. Notably, ARGS demonstrates consistent enhancements in average reward compared to baselines across diverse alignment tasks and various model dimensions. For example, under the same greedy-based decoding strategy, our method improves the average reward by 19.56% relative to the baseline and secures a preference or tie score of 64.33% in GPT-4 evaluation. We believe that our framework, emphasizing decoding-time alignment, paves the way for more responsive language models in the future. Code is publicly available at: \url{https://github.com/deeplearning-wisc/args}.

Lab results are often confusing and hard to understand. Large language models (LLMs) such as ChatGPT have opened a promising avenue for patients to get their questions answered. We aim to assess the feasibility of using LLMs to generate relevant, accurate, helpful, and unharmful responses to lab test-related questions asked by patients and to identify potential issues that can be mitigated with augmentation approaches. We first collected lab test results related question and answer data from Yahoo! Answers and selected 53 QA pairs for this study. Using the LangChain framework and ChatGPT web portal, we generated responses to the 53 questions from four LLMs including GPT-4, Meta LLaMA 2, MedAlpaca, and ORCA_mini. We first assessed the similarity of their answers using standard QA similarity-based evaluation metrics including ROUGE, BLEU, METEOR, BERTScore. We also utilized an LLM-based evaluator to judge whether a target model has higher quality in terms of relevance, correctness, helpfulness, and safety than the baseline model. Finally, we performed a manual evaluation with medical experts for all the responses to seven selected questions on the same four aspects. The results of Win Rate and medical expert evaluation both showed that GPT-4's responses achieved better scores than all the other LLM responses and human responses on all four aspects (relevance, correctness, helpfulness, and safety). However, LLM responses occasionally also suffer from a lack of interpretation in one's medical context, incorrect statements, and lack of references. We find that compared to other three LLMs and human answer from the Q&A website, GPT-4's responses are more accurate, helpful, relevant, and safer. However, there are cases which GPT-4 responses are inaccurate and not individualized. We identified a number of ways to improve the quality of LLM responses.

However, as of the current state of research, there appears to be a significant gap in exploratory and experimental studies specifically addressing the capabilities of Generative AI and Large Language Models (LLMs) in the context of determination and discovery of legal issues. Such studies would be instrumental in understanding how these advanced AI technologies manage the intricate task of accurately classifying and pinpointing legal matters, a domain traditionally reliant on the deep, contextual, and specialised knowledge of human legal experts. To address the identified gap in the research landscape, this study proposes an experimental and exploratory analysis of the performance of LLMs in the legal domain. The research aims to evaluate the capabilities of LLMs contrasting their performance against human legal practitioners on high volume real-world legal tasks. These types of high volume legal tasks are frequently outsourced or pushed to less experienced lawyers, and given the rapid advancements made by LLMs, raises the question of whether LLMs have achieved a level of legal comprehension that is comparable to the quality, accuracy and efficiency of Junior Lawyers or outsourced legal practitioners on such tasks.

A major challenge in Natural Language Processing is obtaining annotated data for supervised learning. An option is the use of crowdsourcing platforms for data annotation. However, crowdsourcing introduces issues related to the annotator's experience, consistency, and biases. An alternative is to use zero-shot methods, which in turn have limitations compared to their few-shot or fully supervised counterparts. Recent advancements driven by large language models show potential, but struggle to adapt to specialized domains with severely limited data. The most common approaches therefore involve the human itself randomly annotating a set of datapoints to build initial datasets. But randomly sampling data to be annotated is often inefficient as it ignores the characteristics of the data and the specific needs of the model. The situation worsens when working with imbalanced datasets, as random sampling tends to heavily bias towards the majority classes, leading to excessive annotated data. To address these issues, this paper contributes an automatic and informed data selection architecture to build a small dataset for few-shot learning. Our proposal minimizes the quantity and maximizes diversity of data selected for human annotation, while improving model performance.

In this work, we address question answering (QA) over a hybrid of tabular and textual data that are very common content on the Web (e.g. SEC filings), where discrete reasoning capabilities are often required. Recently, large language models (LLMs) like GPT-4 have demonstrated strong multi-step reasoning capabilities. We then consider harnessing the amazing power of LLMs to solve our task. We abstract a Step-wise Pipeline for tabular and textual QA, which consists of three key steps, including Extractor, Reasoner and Executor, and initially design an instruction to instantiate the pipeline and validate that GPT-4 outperforms all existing methods. However, utilizing an online LLM like GPT-4 holds various challenges in terms of cost, latency, and data security risk, which motivates us to specialize smaller LLMs in this task. We develop a TAT-LLM language model by fine-tuning LLaMA 2 with the training data generated automatically from existing expert-annotated datasets following the Step-wise Pipeline. The experimental results have verified that our TAT-LLM model can outperform all baseline models, including the previous best fine-tuned models and very large-scale LLMs like GPT-4 on FinQA, TAT-QA and TAT-DQA benchmarks. We hope our work can serve as a pioneering example of specializing smaller language models for specific tasks.

A species' genetic code or genome encodes valuable evolutionary, biological, and phylogenetic information that aids in species recognition, taxonomic classification, and understanding genetic predispositions like drug resistance and virulence. However, the vast number of potential species poses significant challenges in developing a general-purpose whole genome classification tool. Traditional bioinformatics tools have made notable progress but lack scalability and are computationally expensive. Machine learning-based frameworks show promise but must address the issue of large classification vocabularies with long-tail distributions. In this study, we propose addressing this problem through zero-shot learning using TEPI, Taxonomy-aware Embedding and Pseudo-Imaging. We represent each genome as pseudo-images and map them to a taxonomy-aware embedding space for reasoning and classification. This embedding space captures compositional and phylogenetic relationships of species, enabling predictions in extensive search spaces. We evaluate TEPI using two rigorous zero-shot settings and demonstrate its generalization capabilities qualitatively on curated, large-scale, publicly sourced data.

Structural extraction of events within discourse is critical since it avails a deeper understanding of communication patterns and behavior trends. Event argument extraction (EAE), at the core of event-centric understanding, is the task of identifying role-specific text spans (i.e., arguments) for a given event. Document-level EAE (DocEAE) focuses on arguments that are scattered across an entire document. In this work, we explore the capabilities of open source Large Language Models (LLMs), i.e., Flan-UL2, for the DocEAE task. To this end, we propose ULTRA, a hierarchical framework that extracts event arguments more cost-effectively -- the method needs as few as 50 annotations and doesn't require hitting costly API endpoints. Further, it alleviates the positional bias issue intrinsic to LLMs. ULTRA first sequentially reads text chunks of a document to generate a candidate argument set, upon which ULTRA learns to drop non-pertinent candidates through self-refinement. We further introduce LEAFER to address the challenge LLMs face in locating the exact boundary of an argument span. ULTRA outperforms strong baselines, which include strong supervised models and ChatGPT, by 9.8% when evaluated by the exact match (EM) metric.