Large Language Models (LLMs) are smart but forgetful. Recent studies, (e.g., (Bubeck et al., 2023)) on modern LLMs have shown that they are capable of performing amazing tasks typically necessitating human-level intelligence. However, unlike humans, frozen LLMs do not improve over time; they neither acquire new knowledge nor learn from their successes or failures. Some approaches to improving the intelligence of LLMs include fine-tuning models based on problem-solving performance (Zelikman et al., 2022), and building bigger and more sophisticated models (Bubeck et al., 2023). However, these methods have the drawback of requiring substantial data and computational resources to retrain existing models. In this paper, we explore the use of Retrieval Augmented Generation, also known as RAG (Lewis et al., 2021) to improve problem-solving performance. We propose ARM-RAG (Auxiliary Rationale Memory for Retrieval Augmented Generation), a system that learns from its successes without incurring high training costs. We demonstrate that the storage and subsequent retrieval of reasoning chains have a positive influence on performance in grade-school math problems.

An explosion of work in language is leading to ever-increasing numbers of available natural language processing models, with little understanding of how new models compare to better-understood models. One major reason for this difficulty is saturating benchmark datasets, which may not reflect well differences in model performance in the wild. In this work, we propose a novel framework for comparing two natural language processing models by revealing their shared invariance to interpretable input perturbations that are designed to target a specific linguistic capability (e.g., Synonym-Invariance, Typo-Invariance). Via experiments on models from within the same and across different architecture families, this framework offers a number of insights about how changes in models (e.g., distillation, increase in size, amount of pre-training) affect multiple well-defined linguistic capabilities. Furthermore, we also demonstrate how our framework can enable evaluation of the invariances shared between models that are available as commercial black-box APIs (e.g., InstructGPT family) and models that are relatively better understood (e.g., GPT-2). Across several experiments, we observe that large language models share many of the invariances encoded by models of various sizes, whereas the invariances encoded by large language models are only shared by other large models. Possessing a wide variety of invariances may be a key reason for the recent successes of large language models, and our framework can shed light on the types of invariances that are retained by or emerge in new models. A key reason for the tremendous progress and adoption of natural language processing (NLP) models has been the ready availability of models that can be effectively adapted to diverse downstream tasks and datasets (Wolf et al., 2019). However, with the increasing number of new models, it is difficult to know how new models compare to better-understood ones. This is complicated by the fact that standard benchmark datasets are saturating (Dehghani et al., 2021; Owen, 2023), and small differences on these datasets may in fact correspond to large differences in model performance in the wild (Tay et al., 2022; Zhang et al., 2022; Liu et al., 2023). To enable more comprehensive model comparisons, we propose a novel framework for comparing two natural language processing models by investigating their shared invariance to specific input perturbations. We focus specifically on evaluating invariances that are shared between models, as the invariances encoded by a model implicitly define the features of data that the model deems important and is consequently sensitive to, as well as delineate the features it finds irrelevant.

This paper provides different approaches for a binary sentiment classification on a small training dataset. LLMs that provided state-of-the-art results in sentiment analysis and similar domains are being used, such as BERT, RoBERTa and XLNet. The reader can also find different data augmentation techniques to deal with the small amount of training data provided.

As large language models become more prevalent, their possible harmful or inappropriate responses are a cause for concern. This paper introduces a unique dataset containing adversarial examples in the form of questions, which we call AttaQ, designed to provoke such harmful or inappropriate responses. We assess the efficacy of our dataset by analyzing the vulnerabilities of various models when subjected to it. Additionally, we introduce a novel automatic approach for identifying and naming vulnerable semantic regions - input semantic areas for which the model is likely to produce harmful outputs. This is achieved through the application of specialized clustering techniques that consider both the semantic similarity of the input attacks and the harmfulness of the model's responses. Automatically identifying vulnerable semantic regions enhances the evaluation of model weaknesses, facilitating targeted improvements to its safety mechanisms and overall reliability.

Alignment with human preference is a desired property of large language models (LLMs). Currently, the main alignment approach is based on reinforcement learning from human feedback (RLHF). Despite the effectiveness of RLHF, it is intricate to implement and train, thus recent studies explore how to develop alternative alignment approaches based on supervised fine-tuning (SFT). A major limitation of SFT is that it essentially does imitation learning, which cannot fully understand what are the expected behaviors. To address this issue, we propose an improved alignment approach named FIGA. Different from prior methods, we incorporate fine-grained (i.e., token or phrase level) quality signals that are derived by contrasting good and bad responses. Our approach has made two major contributions. Firstly, we curate a refined alignment dataset that pairs initial responses and the corresponding revised ones. Secondly, we devise a new loss function can leverage fine-grained quality signals to instruct the learning of LLMs for alignment. Extensive experiments have demonstrated the effectiveness of our approaches by comparing a number of competitive baselines.

Interactive and embodied tasks pose at least two fundamental challenges to existing Vision & Language (VL) models, including 1) grounding language in trajectories of actions and observations, and 2) referential disambiguation. To tackle these challenges, we propose an Embodied MultiModal Agent (EMMA): a unified encoder-decoder model that reasons over images and trajectories, and casts action prediction as multimodal text generation. By unifying all tasks as text generation, EMMA learns a language of actions which facilitates transfer across tasks. Different to previous modular approaches with independently trained components, we use a single multitask model where each task contributes to goal completion. EMMA performs on par with similar models on several VL benchmarks and sets a new state-of-the-art performance (36.81% success rate) on the Dialog-guided Task Completion (DTC), a benchmark to evaluate dialog-guided agents in the Alexa Arena

Explainable Artificial Intelligence (XAI) is an emerging field in AI that aims to address the opaque nature of machine learning models. Furthermore, it has been shown that XAI can be used to extract input-output relationships, making them a useful tool in chemistry to understand structure-property relationships. However, one of the main limitations of XAI methods is that they are developed for technically oriented users. We propose the XpertAI framework that integrates XAI methods with large language models (LLMs) accessing scientific literature to generate accessible natural language explanations of raw chemical data automatically. We conducted 5 case studies to evaluate the performance of XpertAI. Our results show that XpertAI combines the strengths of LLMs and XAI tools in generating specific, scientific, and interpretable explanations.

Many capable large language models (LLMs) are developed via self-supervised pre-training followed by a reinforcement-learning fine-tuning phase, often based on human or AI feedback. During this stage, models may be guided by their inductive biases to rely on simpler features which may be easier to extract, at a cost to robustness and generalisation. We investigate whether principles governing inductive biases in the supervised fine-tuning of LLMs also apply when the fine-tuning process uses reinforcement learning. Following Lovering et al (2021), we test two hypotheses: that features more $\textit{extractable}$ after pre-training are more likely to be utilised by the final policy, and that the evidence for/against a feature predicts whether it will be utilised. Through controlled experiments on synthetic and natural language tasks, we find statistically significant correlations which constitute strong evidence for these hypotheses.

The rapid development of language models (LMs) brings unprecedented accessibility and usage for both models and users. On the one hand, powerful LMs, trained with massive textual data, achieve state-of-the-art performance over numerous downstream NLP tasks. On the other hand, more and more attention is paid to unrestricted model accesses that may bring malicious privacy risks of data leakage. To address these issues, many recent works propose privacy-preserving language models (PPLMs) with differential privacy (DP). Unfortunately, different DP implementations make it challenging for a fair comparison among existing PPLMs. In this paper, we present P-Bench, a multi-perspective privacy evaluation benchmark to empirically and intuitively quantify the privacy leakage of LMs. Instead of only protecting and measuring the privacy of protected data with DP parameters, P-Bench sheds light on the neglected inference data privacy during actual usage. P-Bench first clearly defines multi-faceted privacy objectives during private fine-tuning. Then, P-Bench constructs a unified pipeline to perform private fine-tuning. Lastly, P-Bench performs existing privacy attacks on LMs with pre-defined privacy objectives as the empirical evaluation results. The empirical attack results are used to fairly and intuitively evaluate the privacy leakage of various PPLMs. We conduct extensive experiments on three datasets of GLUE for mainstream LMs.

Generative artificial intelligence (GenAI) offers promising potential for advancing human-AI collaboration in qualitative research. However, existing works focused on conventional machine-learning and pattern-based AI systems, and little is known about how researchers interact with GenAI in qualitative research. This work delves into researchers' perceptions of their collaboration with GenAI, specifically ChatGPT. Through a user study involving ten qualitative researchers, we found ChatGPT to be a valuable collaborator for thematic analysis, enhancing coding efficiency, aiding initial data exploration, offering granular quantitative insights, and assisting comprehension for non-native speakers and non-experts. Yet, concerns about its trustworthiness and accuracy, reliability and consistency, limited contextual understanding, and broader acceptance within the research community persist. We contribute five actionable design recommendations to foster effective human-AI collaboration. These include incorporating transparent explanatory mechanisms, enhancing interface and integration capabilities, prioritising contextual understanding and customisation, embedding human-AI feedback loops and iterative functionality, and strengthening trust through validation mechanisms.