Large Language Models (LLMs) exhibit substantial capabilities yet encounter challenges, including hallucination, outdated knowledge, and untraceable reasoning processes. Retrieval-augmented generation (RAG) has emerged as a promising solution, integrating knowledge from external databases to mitigate these challenges. However, inappropriate retrieved passages can potentially hinder the LLMs' capacity to generate comprehensive and high-quality responses. Prior RAG studies on the robustness of retrieval noises often confine themselves to a limited set of noise types, deviating from real-world retrieval environments and limiting practical applicability. In this study, we initially investigate retrieval noises and categorize them into three distinct types, reflecting real-world environments. We analyze the impact of these various retrieval noises on the robustness of LLMs. Subsequently, we propose a novel RAG approach known as Retrieval-augmented Adaptive Adversarial Training (RAAT). RAAT leverages adaptive adversarial training to dynamically adjust the model's training process in response to retrieval noises. Concurrently, it employs multi-task learning to ensure the model's capacity to internally recognize noisy contexts. Extensive experiments demonstrate that the LLaMA-2 7B model trained using RAAT exhibits significant improvements in F1 and EM scores under diverse noise conditions. For reproducibility, we release our code and data at: https://github.com/calubkk/RAAT.

Generative AI offers significant opportunities for language learning. Tools like ChatGPT can provide informal second language practice through chats in written or voice forms, with the learner specifying through prompts conversational parameters such as proficiency level, language register, and discussion topics. AI can be instructed to give corrective feedback, create practice exercises, or develop an extended study plan. Instructors can use AI to build learning and assessment materials in a variety of media. AI is likely to make immersive technologies more powerful and versatile, moving away from scripted interactions. For both learners and teachers, it is important to understand the limitations of AI systems that arise from their purely statistical model of human language, which limits their ability to deal with nuanced social and cultural aspects of language use. Additionally, there are ethical concerns over how AI systems are created as well as practical constraints in their use, especially for less privileged populations. The power and versatility of AI tools are likely to turn them into valuable and constant companions in many peoples lives (akin to smartphones), creating a close connection that goes beyond simple tool use. Ecological theories such as sociomaterialism are helpful in examining the shared agency that develops through close user-AI interactions, as are the perspectives on human-object relations from Indigenous cultures.

The scarcity of non-English data limits the development of non-English large language models (LLMs). Transforming English-centric LLMs to non-English has been identified as an effective and resource-efficient method. Previous works start from base LLMs and perform knowledge distillation (KD) with data generated by stronger LLMs, e.g. GPT-4. Compared to base LLMs, chat LLMs are further optimized for advanced abilities, e.g. multi-turn conversation and human preference alignment, and thus more powerful in both helpfulness and safety. However, transforming a chat LLM involves two critical issues: (1) How can we effectively transfer advanced abilities without their supervised data? (2) How can we prevent the original knowledge from catastrophic forgetting during transformation? We target these issues by introducing a simple framework called TransLLM. For the first issue, TransLLM divides the transfer problem into some common sub-tasks with the translation chain-of-thought, which uses the translation as the bridge between English and non-English step-by-step. We further enhance the performance of sub-tasks with publicly available data. For the second issue, we propose a method comprising two synergistic components: low-rank adaptation for training to maintain the original LLM parameters, and recovery KD, which utilizes data generated by the chat LLM itself to recover the original knowledge from the frozen parameters. In the experiments, we transform the LLaMA-2-chat-7B to the Thai language. Our method, using only single-turn data, outperforms strong baselines and ChatGPT on multi-turn benchmark MT-bench. Furthermore, our method, without safety data, rejects more harmful queries of safety benchmark AdvBench than both ChatGPT and GPT-4.

One uniquely human trait is our inability to be random. We see and produce patterns where there should not be any and we do so in a predictable way. LLMs are supplied with human data and prone to human biases. In this work, we explore how LLMs approach randomness and where and how they fail through the lens of the well studied phenomena of generating binary random sequences. We find that GPT 4 and Llama 3 exhibit and exacerbate nearly every human bias we test in this context, but GPT 3.5 exhibits more random behavior. This dichotomy of randomness or humaness is proposed as a fundamental question of LLMs and that either behavior may be useful in different circumstances.

Multimodal large language models (MLLMs) contribute a powerful mechanism to understanding visual information building on large language models. However, MLLMs are notorious for suffering from hallucinations, especially when generating lengthy, detailed descriptions for images. Our analysis reveals that hallucinations stem from the inherent summarization mechanism of large language models, leading to excessive dependence on linguistic tokens while neglecting vision information. In this paper, we propose NoiseBoost, a broadly applicable and simple method for alleviating hallucinations for MLLMs through the integration of noise feature perturbations. Noise perturbation acts as a regularizer, facilitating a balanced distribution of attention weights among visual and linguistic tokens. Despite its simplicity, NoiseBoost consistently enhances the performance of MLLMs across common training strategies, including supervised fine-tuning and reinforcement learning. Further, NoiseBoost pioneerly enables semi-supervised learning for MLLMs, unleashing the power of unlabeled data. Comprehensive experiments demonstrate that NoiseBoost improves dense caption accuracy by 8.1% with human evaluation and achieves comparable results with 50% of the data by mining unlabeled data. Code and models are available at https://kaiwu5.github.io/noiseboost.

Reinforcement Learning from Human Feedback (RLHF) has been commonly used to align the behaviors of Large Language Models (LLMs) with human preferences. Recently, a popular alternative is Direct Policy Optimization (DPO), which replaces an LLM-based reward model with the policy itself, thus obviating the need for extra memory and training time to learn the reward model. However, DPO does not consider the relative qualities of the positive and negative responses, and can lead to sub-optimal training outcomes. To alleviate this problem, we investigate the use of intrinsic knowledge within the on-the-fly fine-tuning LLM to obtain relative qualities and help to refine the loss function. Specifically, we leverage the knowledge of the LLM to design a refinement function to estimate the quality of both the positive and negative responses. We show that the constructed refinement function can help self-refine the loss function under mild assumptions. The refinement function is integrated into DPO and its variant Identity Policy Optimization (IPO). Experiments across various evaluators indicate that they can improve the performance of the fine-tuned models over DPO and IPO.

Large language models (LLMs) showcase impressive reasoning capabilities when coupled with Chain-of-Thought (CoT) prompting. However, the robustness of this approach warrants further investigation. In this paper, we introduce a novel scenario termed preemptive answers, where the LLM obtains an answer before engaging in reasoning. This situation can arise inadvertently or induced by malicious users by prompt injection attacks. Experiments reveal that preemptive answers significantly impair the model's reasoning capability across various CoT methods and a broad spectrum of datasets. To bolster the robustness of reasoning, we propose two measures aimed at mitigating this issue to some extent.

First, effectiveness of their utterances when faced with we extend the investigation to a much larger corpus multiple options for structuring a message. The of informal written English collected from social UID hypothesis (Frank and Jaeger, 2008; Collins, media. Second, we use contemporary large language 2014; Hahn et al., 2020) suggests that speakers models (LLMs) to estimate the operationalizations tend to spread information evenly throughout an of information uniformity in syntactic reduction, utterance, avoiding large fluctuations in the perunit suggesting the robustness of our findings.

As large language models (LLMs) continue to demonstrate remarkable abilities across various domains, computer scientists are developing methods to understand their cognitive processes, particularly concerning how (and if) LLMs internally represent their beliefs about the world. However, this field currently lacks a unified theoretical foundation to underpin the study of belief in LLMs. This article begins filling this gap by proposing adequacy conditions for a representation in an LLM to count as belief-like. We argue that, while the project of belief measurement in LLMs shares striking features with belief measurement as carried out in decision theory and formal epistemology, it also differs in ways that should change how we measure belief. Thus, drawing from insights in philosophy and contemporary practices of machine learning, we establish four criteria that balance theoretical considerations with practical constraints. Our proposed criteria include accuracy, coherence, uniformity, and use, which together help lay the groundwork for a comprehensive understanding of belief representation in LLMs. We draw on empirical work showing the limitations of using various criteria in isolation to identify belief representations.

This technical report introduces Aya 23, a family of multilingual language models. Aya 23 builds on the recent release of the Aya model (\"Ust\"un et al., 2024), focusing on pairing a highly performant pre-trained model with the recently released Aya collection (Singh et al., 2024). The result is a powerful multilingual large language model serving 23 languages, expanding state-of-art language modeling capabilities to approximately half of the world's population. The Aya model covered 101 languages whereas Aya 23 is an experiment in depth vs breadth, exploring the impact of allocating more capacity to fewer languages that are included during pre-training. Aya 23 outperforms both previous massively multilingual models like Aya 101 for the languages it covers, as well as widely used models like Gemma, Mistral and Mixtral on an extensive range of discriminative and generative tasks. We release the open weights for both the 8B and 35B models as part of our continued commitment for expanding access to multilingual progress.