This paper presents the development of a novel ethical reasoning framework for robots. "Robots Can Feel" is the first system for robots that utilizes a combination of logic and human-like emotion simulation to make decisions in morally complex situations akin to humans. The key feature of the approach is the management of the Emotion Weight Coefficient - a customizable parameter to assign the role of emotions in robot decision-making. The system aims to serve as a tool that can equip robots of any form and purpose with ethical behavior close to human standards. Besides the platform, the system is independent of the choice of the base model. During the evaluation, the system was tested on 8 top up-to-date LLMs (Large Language Models). This list included both commercial and open-source models developed by various companies and countries. The research demonstrated that regardless of the model choice, the Emotions Weight Coefficient influences the robot's decision similarly. According to ANOVA analysis, the use of different Emotion Weight Coefficients influenced the final decision in a range of situations, such as in a request for a dietary violation F(4, 35) = 11.2, p = 0.0001 and in an animal compassion situation F(4, 35) = 8.5441, p = 0.0001. A demonstration code repository is provided at: https://github.com/TemaLykov/robots_can_feel

Combining different forms of prompts with pre-trained large language models has yielded remarkable results on reasoning tasks (e.g. Chain-of-Thought prompting). However, along with testing on more complex reasoning, these methods also expose problems such as invalid reasoning and fictional reasoning paths. In this paper, we develop \textit{Hypothesis Testing Prompting}, which adds conclusion assumptions, backward reasoning, and fact verification during intermediate reasoning steps. \textit{Hypothesis Testing prompting} involves multiple assumptions and reverses validation of conclusions leading to its unique correct answer. Experiments on two challenging deductive reasoning datasets ProofWriter and RuleTaker show that hypothesis testing prompting not only significantly improves the effect, but also generates a more reasonable and standardized reasoning process.

Commonsense question answering has demonstrated considerable potential across various applications like assistants and social robots. Although fully fine-tuned pre-trained Language Models(LM) have achieved remarkable performance in commonsense reasoning, their tendency to excessively prioritize textual information hampers the precise transfer of structural knowledge and undermines interpretability. Some studies have explored combining LMs with Knowledge Graphs(KGs) by coarsely fusing the two modalities to perform Graph Neural Network(GNN)-based reasoning that lacks a profound interaction between heterogeneous modalities. In this paper, we propose a novel Graph-based Structure-Aware Prompt Learning Model for commonsense reasoning, named G-SAP, aiming to maintain a balance between heterogeneous knowledge and enhance the cross-modal interaction within the LM+GNNs model. In particular, an evidence graph is constructed by integrating multiple knowledge sources, i.e. ConceptNet, Wikipedia, and Cambridge Dictionary to boost the performance. Afterward, a structure-aware frozen PLM is employed to fully incorporate the structured and textual information from the evidence graph, where the generation of prompts is driven by graph entities and relations. Finally, a heterogeneous message-passing reasoning module is used to facilitate deep interaction of knowledge between the LM and graph-based networks. Empirical validation, conducted through extensive experiments on three benchmark datasets, demonstrates the notable performance of the proposed model. The results reveal a significant advancement over the existing models, especially, with 6.12% improvement over the SoTA LM+GNNs model on the OpenbookQA dataset.

Qualitative analysis is a challenging, yet crucial aspect of advancing research in the field of Human-Computer Interaction (HCI). Recent studies show that large language models (LLMs) can perform qualitative coding within existing schemes, but their potential for collaborative human-LLM discovery and new insight generation in qualitative analysis is still underexplored. To bridge this gap and advance qualitative analysis by harnessing the power of LLMs, we propose CHALET, a novel methodology that leverages the human-LLM collaboration paradigm to facilitate conceptualization and empower qualitative research. The CHALET approach involves LLM-supported data collection, performing both human and LLM deductive coding to identify disagreements, and performing collaborative inductive coding on these disagreement cases to derive new conceptual insights. We validated the effectiveness of CHALET through its application to the attribution model of mental-illness stigma, uncovering implicit stigmatization themes on cognitive, emotional and behavioral dimensions. We discuss the implications for future research, methodology, and the transdisciplinary opportunities CHALET presents for the HCI community and beyond.

Large Language Models (LLMs) have demonstrated their efficacy across a broad spectrum of tasks in healthcare applications. However, often LLMs need to be fine-tuned on taskspecific expert-annotated data to achieve optimal performance, which can be expensive and time consuming. In this study, we fine-tune PaLM-2 (Anil et al. (2023)) with parameter efficient finetuning (PEFT) using noisy labels obtained from gemini-pro 1.0 (Google (2024)) for the detection of Schedule-of-Event (SoE) tables, which specify care plan in clinical trial protocols. We introduce a filtering mechanism to select high-confidence labels for this table classification task, thereby reducing the noise in the auto-generated labels. We show that fine-tuned PaLM-2 with those labels achieves performance that exceeds the gemini-pro 1.0 and other LLMs. Furthermore, its performance is close to a PaLM-2 fine-tuned on labels obtained from non-expert annotators. Our results show that leveraging LLM-generated labels through powerful models like gemini-pro can potentially serve as a viable strategy for improving LLM performance through fine-tuning in specialized tasks, particularly in domains where expert annotations are scarce, expensive, or time-consuming to obtain.

Multimodal large language models (MLLMs) demand considerable computations for inference due to the extensive parameters and the additional input tokens needed for visual information representation. Herein, we introduce Visual Tokens Withdrawal (VTW), a plug-and-play module to boost MLLMs for rapid inference. Our approach is inspired by two intriguing phenomena we have observed: (1) the attention sink phenomenon that is prevalent in LLMs also persists in MLLMs, suggesting that initial tokens and nearest tokens receive the majority of attention, while middle vision tokens garner minimal attention in deep layers; (2) the presence of information migration, which implies that visual information is transferred to subsequent text tokens within the first few layers of MLLMs. As per our findings, we conclude that vision tokens are not necessary in the deep layers of MLLMs. Thus, we strategically withdraw them at a certain layer, enabling only text tokens to engage in subsequent layers. To pinpoint the ideal layer for vision tokens withdrawal, we initially analyze a limited set of tiny datasets and choose the first layer that meets the Kullback-Leibler divergence criterion. Our VTW approach can cut computational overhead by over 40\% across diverse multimodal tasks while maintaining performance. Our code is released at https://github.com/lzhxmu/VTW.

With an evergrowing number of LLMs reporting superlative performance for English, their ability to perform equitably for different dialects of English (i.e., dialect robustness) needs to be ascertained. Specifically, we use English language (US English or Indian English) conversations between humans who play the word-guessing game of `taboo'. We formulate two evaluative tasks: target word prediction (TWP) (i.e.predict the masked target word in a conversation) and target word selection (TWS) (i.e., select the most likely masked target word in a conversation, from among a set of candidate words). Extending MD3, an existing dialectic dataset of taboo-playing conversations, we introduce M-MD3, a target-word-masked version of MD3 with the USEng and IndEng subsets. We add two subsets: AITrans (where dialectic information is removed from IndEng) and AIGen (where LLMs are prompted to generate conversations). Our evaluation uses pre-trained and fine-tuned versions of two closed-source (GPT-4/3.5) and two open-source LLMs (Mistral and Gemma). LLMs perform significantly better for US English than Indian English for both TWP and TWS, for all settings. While GPT-based models perform the best, the comparatively smaller models work more equitably for short conversations (<8 turns). Our results on AIGen and AITrans (the best and worst-performing subset) respectively show that LLMs may learn a dialect of their own based on the composition of the training data, and that dialect robustness is indeed a challenging task. Our evaluation methodology exhibits a novel way to examine attributes of language models using pre-existing dialogue datasets.

Abstract-- This research introduces the Bi-VLA (Vision-Language-Action) model, a novel system designed for bimanual robotic dexterous manipulation that seamlessly integrates vision for scene understanding, language comprehension for translating human instructions into executable code, and physical action generation. We evaluated the system's functionality through a series of household tasks, including the preparation of a desired salad upon human request. Bi-VLA demonstrates the ability to interpret complex human instructions, perceive and understand the visual context of ingredients, and execute precise bimanual actions to prepare the requested salad. We assessed the system's performance in terms of accuracy, efficiency, and adaptability to different salad recipes and human preferences through a series of experiments. Our results show a 100% success rate in generating the correct executable code by the Language Module, a 96.06% success rate in detecting specific ingredients by the Vision Module, and an overall success rate of 83.4% in However, despite their potential, the application of language models Recent advancements in language models have significantly to synthesize the bimanual skills of robots has not received impacted Human-Robot Interaction (HRI), enabling significant attention.

Recent studies have shown that Large Language Models (LLMs) have the potential to process extremely long text. Many works only evaluate LLMs' long-text processing ability on the language modeling task, with perplexity (PPL) as the evaluation metric. However, in our study, we find that there is no correlation between PPL and LLMs' long-text understanding ability. Besides, PPL may only reflect the model's ability to model local information instead of catching long-range dependency. Therefore, only using PPL to prove the model could process long text is inappropriate. The local focus feature of PPL could also explain some existing phenomena, such as the great extrapolation ability of the position method ALiBi. When evaluating a model's ability in long text, we might pay more attention to PPL's limitation and avoid overly relying on it.

While large multi-modal models (LMM) have shown notable progress in multi-modal tasks, their capabilities in tasks involving dense textual content remains to be fully explored. Dense text, which carries important information, is often found in documents, tables, and product descriptions. Understanding dense text enables us to obtain more accurate information, assisting in making better decisions. To further explore the capabilities of LMM in complex text tasks, we propose the DT-VQA dataset, with 170k question-answer pairs. In this paper, we conduct a comprehensive evaluation of GPT4V, Gemini, and various open-source LMMs on our dataset, revealing their strengths and weaknesses. Furthermore, we evaluate the effectiveness of two strategies for LMM: prompt engineering and downstream fine-tuning. We find that even with automatically labeled training datasets, significant improvements in model performance can be achieved. We hope that this research will promote the study of LMM in dense text tasks.