The widespread application of Large Language Models (LLMs) across various tasks and fields has necessitated the alignment of these models with human values and preferences. Given various approaches of human value alignment, ranging from Reinforcement Learning with Human Feedback (RLHF), to constitutional learning, etc. there is an urgent need to understand the scope and nature of human values injected into these models before their release. There is also a need for model alignment without a costly large scale human annotation effort. We propose UniVaR, a high-dimensional representation of human value distributions in LLMs, orthogonal to model architecture and training data. Trained from the value-relevant output of eight multilingual LLMs and tested on the output from four multilingual LLMs, namely LlaMA2, ChatGPT, JAIS and Yi, we show that UniVaR is a powerful tool to compare the distribution of human values embedded in different LLMs with different langauge sources. Through UniVaR, we explore how different LLMs prioritize various values in different languages and cultures, shedding light on the complex interplay between human values and language modeling.
Inference, especially those derived from inductive processes, is a crucial component in our conversation to complement the information implicitly or explicitly conveyed by a speaker. While recent large language models show remarkable advances in inference tasks, their performance in inductive reasoning, where not all information is present in the context, is far behind deductive reasoning. In this paper, we analyze the behavior of the models based on the task difficulty defined by the semantic information gap -- which distinguishes inductive and deductive reasoning (Johnson-Laird, 1988, 1993). Our analysis reveals that the disparity in information between dialogue contexts and desired inferences poses a significant challenge to the inductive inference process. To mitigate this information gap, we investigate a contrastive learning approach by feeding negative samples. Our experiments suggest negative samples help models understand what is wrong and improve their inference generations.
Large language models (LLMs) have shown promise for generative and knowledge-intensive tasks including question-answering (QA) tasks. However, the practical deployment still faces challenges, notably the issue of "hallucination", where models generate plausible-sounding but unfaithful or nonsensical information. This issue becomes particularly critical in the medical domain due to the uncommon professional concepts and potential social risks involved. This paper analyses the phenomenon of hallucination in medical generative QA systems using widely adopted LLMs and datasets. Our investigation centers on the identification and comprehension of common problematic answers, with a specific emphasis on hallucination. To tackle this challenge, we present an interactive self-reflection methodology that incorporates knowledge acquisition and answer generation. Through this feedback process, our approach steadily enhances the factuality, consistency, and entailment of the generated answers. Consequently, we harness the interactivity and multitasking ability of LLMs and produce progressively more precise and accurate answers. Experimental results on both automatic and human evaluation demonstrate the superiority of our approach in hallucination reduction compared to baselines.
Question rewriting (QR) is a subtask of conversational question answering (CQA) aiming to ease the challenges of understanding dependencies among dialogue history by reformulating questions in a self-contained form. Despite seeming plausible, little evidence is available to justify QR as a mitigation method for CQA. To verify the effectiveness of QR in CQA, we investigate a reinforcement learning approach that integrates QR and CQA tasks and does not require corresponding QR datasets for targeted CQA. We find, however, that the RL method is on par with the end-to-end baseline. We provide an analysis of the failure and describe the difficulty of exploiting QR for CQA.
Automatic script generation could save a considerable amount of resources and offer inspiration to professional scriptwriters. We present VScript, a controllable pipeline that generates complete scripts including dialogues and scene descriptions, and presents visually using video retrieval and aurally using text-to-speech for spoken dialogue. With an interactive interface, our system allows users to select genres and input starting words that control the theme and development of the generated script. We adopt a hierarchical structure, which generates the plot, then the script and its audio-visual presentation. We also introduce a novel approach to plot-guided dialogue generation by treating it as an inverse dialogue summarization. Experiment results show that our approach outperforms the baselines on both automatic and human evaluations, especially in terms of genre control.
Natural Language Generation (NLG) has improved exponentially in recent years thanks to the development of deep learning technologies such as Transformer-based language models. This advancement has led to more fluent and coherent natural language generation, naturally leading to development in downstream tasks such as abstractive summarization, dialogue generation and data-to-text generation. However, it is also investigated that such generation includes hallucinated texts, which makes the performances of text generation fail to meet users' expectations in many real-world scenarios. In order to address this issue, studies in evaluation and mitigation methods of hallucinations have been presented in various tasks, but have not been reviewed in a combined manner. In this survey, we provide a broad overview of the research progress and challenges in the hallucination problem of NLG. The survey is organized into two big divisions: (i) a general overview of metrics, mitigation methods, and future directions; (ii) task-specific research progress for hallucinations in a large set of downstream tasks: abstractive summarization, dialogue generation, generative question answering, data-to-text generation, and machine translation. This survey could facilitate collaborative efforts among researchers in these tasks.
While the recent advances in deep neural networks (DNN) bring remarkable success, the computational cost also increases considerably. In this paper, we introduce Greenformer, a toolkit to accelerate the computation of neural networks through matrix factorization while maintaining performance. Greenformer can be easily applied with a single line of code to any DNN model. Our experimental results show that Greenformer is effective for a wide range of scenarios. We provide the showcase of Greenformer at https://samuelcahyawijaya.github.io/greenformer-demo/.
Politically sensitive topics are still a challenge for open-domain chatbots. However, dealing with politically sensitive content in a responsible, non-partisan, and safe behavior way is integral for these chatbots. Currently, the main approach to handling political sensitivity is by simply changing such a topic when it is detected. This is safe but evasive and results in a chatbot that is less engaging. In this work, as a first step towards a politically safe chatbot, we propose a group of metrics for assessing their political prudence. We then conduct political prudence analysis of various chatbots and discuss their behavior from multiple angles through our automatic metric and human evaluation metrics. The testsets and codebase are released to promote research in this area.
Information-seeking dialogue systems, including knowledge identification and response generation, aim to respond to users with fluent, coherent, and informative responses based on users' needs, which. To tackle this challenge, we utilize data augmentation methods and several training techniques with the pre-trained language models to learn a general pattern of the task and thus achieve promising performance. In DialDoc21 competition, our system achieved 74.95 F1 score and 60.74 Exact Match score in subtask 1, and 37.72 SacreBLEU score in subtask 2. Empirical analysis is provided to explain the effectiveness of our approaches.