Large Language Models (LLMs) have demonstrated remarkable performance through supervised fine-tuning or in-context learning using gold labels. However, this paradigm is limited by the availability of gold labels, while in certain scenarios, LLMs may need to perform tasks that are too complex for humans to provide such labels. To tackle this challenge, this study explores whether solely utilizing unlabeled data can elicit strong model capabilities. We propose a new paradigm termed zero-to-strong generalization. We iteratively prompt LLMs to annotate unlabeled data and retain high-quality labels by filtering. Surprisingly, we obverse that this iterative process gradually unlocks LLMs' potential on downstream tasks. Our experiments on extensive classification and reasoning tasks confirm the effectiveness of our proposed framework. Our analysis indicates that this paradigm is effective for both in-context learning and fine-tuning, and for various model sizes.

Contextual question-answering models are susceptible to adversarial perturbations to input context, commonly observed in real-world scenarios. These adversarial noises are designed to degrade the performance of the model by distorting the textual input. We introduce a unique dataset that incorporates seven distinct types of adversarial noise into the context, each applied at five different intensity levels on the SQuAD dataset. To quantify the robustness, we utilize robustness metrics providing a standardized measure for assessing model performance across varying noise types and levels. Experiments on transformer-based question-answering models reveal robustness vulnerabilities and important insights into the model's performance in realistic textual input.

Large Language Models (LLMs) have gained popularity in task planning for long-horizon manipulation tasks. To enhance the validity of LLM-generated plans, visual demonstrations and online videos have been widely employed to guide the planning process. However, for manipulation tasks involving subtle movements but rich contact interactions, visual perception alone may be insufficient for the LLM to fully interpret the demonstration. Additionally, visual data provides limited information on force-related parameters and conditions, which are crucial for effective execution on real robots. In this paper, we introduce an in-context learning framework that incorporates tactile and force-torque information from human demonstrations to enhance LLMs' ability to generate plans for new task scenarios. We propose a bootstrapped reasoning pipeline that sequentially integrates each modality into a comprehensive task plan. This task plan is then used as a reference for planning in new task configurations. Real-world experiments on two different sequential manipulation tasks demonstrate the effectiveness of our framework in improving LLMs' understanding of multi-modal demonstrations and enhancing the overall planning performance.

Despite recent advances in training and prompting strategies for Large Language Models (LLMs), these models continue to face challenges with complex logical reasoning tasks that involve long reasoning chains. In this work, we explore the potential and limitations of using graph-based synthetic reasoning data as training signals to enhance LLMs' reasoning capabilities. Our extensive experiments, conducted on two established natural language reasoning tasks -- inductive reasoning and spatial reasoning -- demonstrate that supervised fine-tuning (SFT) with synthetic graph-based reasoning data effectively enhances LLMs' reasoning performance without compromising their effectiveness on other standard evaluation benchmarks.

Large language models (LLMs) have recently emerged as promising tools for solving challenging robotic tasks, even in the presence of action and observation uncertainties. Recent LLM-based decision-making methods (also referred to as LLM-based agents), when paired with appropriate critics, have demonstrated potential in solving complex, long-horizon tasks with relatively few interactions. However, most existing LLM-based agents lack the ability to retain and learn from past interactions - an essential trait of learning-based robotic systems. We propose RAG-Modulo, a framework that enhances LLM-based agents with a memory of past interactions and incorporates critics to evaluate the agents' decisions. The memory component allows the agent to automatically retrieve and incorporate relevant past experiences as in-context examples, providing context-aware feedback for more informed decision-making. Further by updating its memory, the agent improves its performance over time, thereby exhibiting learning. Through experiments in the challenging BabyAI and AlfWorld domains, we demonstrate significant improvements in task success rates and efficiency, showing that the proposed RAG-Modulo framework outperforms state-of-the-art baselines.

Human values and their measurement are long-standing interdisciplinary inquiry. Recent advances in AI have sparked renewed interest in this area, with large language models (LLMs) emerging as both tools and subjects of value measurement. This work introduces Generative Psychometrics for Values (GPV), an LLM-based, data-driven value measurement paradigm, theoretically grounded in text-revealed selective perceptions. We begin by fine-tuning an LLM for accurate perception-level value measurement and verifying the capability of LLMs to parse texts into perceptions, forming the core of the GPV pipeline. Applying GPV to human-authored blogs, we demonstrate its stability, validity, and superiority over prior psychological tools. Then, extending GPV to LLM value measurement, we advance the current art with 1) a psychometric methodology that measures LLM values based on their scalable and free-form outputs, enabling context-specific measurement; 2) a comparative analysis of measurement paradigms, indicating response biases of prior methods; and 3) an attempt to bridge LLM values and their safety, revealing the predictive power of different value systems and the impacts of various values on LLM safety. Through interdisciplinary efforts, we aim to leverage AI for next-generation psychometrics and psychometrics for value-aligned AI.

Understanding emotions is fundamental to human interaction and experience. Humans easily infer emotions from situations or facial expressions, situations from emotions, and do a variety of other affective cognition. How adept is modern AI at these inferences? We introduce an evaluation framework for testing affective cognition in foundation models. Starting from psychological theory, we generate 1,280 diverse scenarios exploring relationships between appraisals, emotions, expressions, and outcomes. We evaluate the abilities of foundation models (GPT-4, Claude-3, Gemini-1.5-Pro) and humans (N = 567) across carefully selected conditions. Our results show foundation models tend to agree with human intuitions, matching or exceeding interparticipant agreement. In some conditions, models are ``superhuman'' -- they better predict modal human judgements than the average human. All models benefit from chain-of-thought reasoning. This suggests foundation models have acquired a human-like understanding of emotions and their influence on beliefs and behavior.

Background: Manual testing is vital for detecting issues missed by automated tests, but specifying accurate verifications is challenging. Aims: This study aims to explore the use of Large Language Models (LLMs) to produce verifications for manual tests. Method: We conducted two independent and complementary exploratory studies. The first study involved using 2 closed-source and 6 open-source LLMs to generate verifications for manual test steps and evaluate their similarity to original verifications. The second study involved recruiting software testing professionals to assess their perception and agreement with the generated verifications compared to the original ones. Results: The open-source models Mistral-7B and Phi-3-mini-4k demonstrated effectiveness and consistency comparable to closed-source models like Gemini-1.5-flash and GPT-3.5-turbo in generating manual test verifications. However, the agreement level among professional testers was slightly above 40%, indicating both promise and room for improvement. While some LLM-generated verifications were considered better than the originals, there were also concerns about AI hallucinations, where verifications significantly deviated from expectations. Conclusion: We contributed by generating a dataset of 37,040 test verifications using 8 different LLMs. Although the models show potential, the relatively modest 40% agreement level highlights the need for further refinement. Enhancing the accuracy, relevance, and clarity of the generated verifications is crucial to ensure greater reliability in real-world testing scenarios.

The introduction of generative pre-trained models, like GPT-4, has introduced a phenomenon known as prompt engineering, whereby model users repeatedly write and revise prompts while trying to achieve a task. Using these AI models for intelligent features in software applications require using APIs that are controlled through developer-written prompts. These prompts have powered AI experiences in popular software products, potentially reaching millions of users. Despite the growing impact of prompt-powered software, little is known about its development process and its relationship to programming. In this work, we argue that some forms of prompts are programs, and that the development of prompts is a distinct phenomenon in programming. We refer to this phenomenon as prompt programming. To this end, we develop an understanding of prompt programming using Straussian grounded theory through interviews with 20 developers engaged in prompt development across a variety of contexts, models, domains, and prompt complexities. Through this study, we contribute 14 observations about prompt programming. For example, rather than building mental models of code, prompt programmers develop mental models of the FM's behavior on the prompt and its unique qualities by interacting with the model. While prior research has shown that experts have well-formed mental models, we find that prompt programmers who have developed dozens of prompts, each with many iterations, still struggle to develop reliable mental models. This contributes to a rapid and unsystematic development process. Taken together, our observations indicate that prompt programming is significantly different from traditional software development, motivating the creation of tools to support prompt programming. Our findings have implications for software engineering practitioners, educators, and researchers.

In this paper, we address the challenge of generating realistic 3D human motions for action classes that were never seen during the training phase. Our approach involves decomposing complex actions into simpler movements, specifically those observed during training, by leveraging the knowledge of human motion contained in GPTs models. These simpler movements are then combined into a single, realistic animation using the properties of diffusion models. Our claim is that this decomposition and subsequent recombination of simple movements can synthesize an animation that accurately represents the complex input action. This method operates during the inference phase and can be integrated with any pre-trained diffusion model, enabling the synthesis of motion classes not present in the training data. We evaluate our method by dividing two benchmark human motion datasets into basic and complex actions, and then compare its performance against the state-of-the-art.