Multimodal Large Language Models (MLLMs) have recently achieved promising zero-shot accuracy on visual question answering (VQA) -- a fundamental task affecting various downstream applications and domains. Given the great potential for the broad use of these models, it is important to investigate their limitations in dealing with different image and question properties. In this work, we investigate whether MLLMs can perceive details as well as larger components in images. In particular, we show that their zero-shot accuracy in answering visual questions is very sensitive to the size of the visual subject related to the question, declining up to $45.91\%$ with size. Furthermore, we show that this effect is causal by observing that human visual cropping can significantly mitigate their sensitivity to size. To scale up the usefulness of human cropping, we propose ViCrop, a general framework that utilizes automatic visual cropping to enhance zero-shot VQA of MLLMs. We construct five variants of ViCrop leveraging either external localization models or the decision process of the given MLLM itself. Our results show that ViCrop improves MLLMs' zero-shot accuracy across different VQA datasets, for example, enhances BLIP2-T5's performance by $32.23\%$ on the TextVQA test set. To facilitate further investigation of MLLMs' behaviors, our code is publicly released.

This paper presents GenH2R, a framework for learning generalizable vision-based human-to-robot (H2R) handover skills. The goal is to equip robots with the ability to reliably receive objects with unseen geometry handed over by humans in various complex trajectories. We acquire such generalizability by learning H2R handover at scale with a comprehensive solution including procedural simulation assets creation, automated demonstration generation, and effective imitation learning. We leverage large-scale 3D model repositories, dexterous grasp generation methods, and curve-based 3D animation to create an H2R handover simulation environment named \simabbns, surpassing the number of scenes in existing simulators by three orders of magnitude. We further introduce a distillation-friendly demonstration generation method that automatically generates a million high-quality demonstrations suitable for learning. Finally, we present a 4D imitation learning method augmented by a future forecasting objective to distill demonstrations into a visuo-motor handover policy. Experimental evaluations in both simulators and the real world demonstrate significant improvements (at least +10\% success rate) over baselines in all cases. The project page is https://GenH2R.github.io/.

The integration of Artificial Intelligence (AI), particularly Large Language Model (LLM)-based systems, in education has shown promise in enhancing teaching and learning experiences. However, the advent of Multimodal Large Language Models (MLLMs) like GPT-4 with vision (GPT-4V), capable of processing multimodal data including text, sound, and visual inputs, opens a new era of enriched, personalized, and interactive learning landscapes in education. Grounded in theory of multimedia learning, this paper explores the transformative role of MLLMs in central aspects of science education by presenting exemplary innovative learning scenarios. Possible applications for MLLMs could range from content creation to tailored support for learning, fostering competencies in scientific practices, and providing assessment and feedback. These scenarios are not limited to text-based and uni-modal formats but can be multimodal, increasing thus personalization, accessibility, and potential learning effectiveness. Besides many opportunities, challenges such as data protection and ethical considerations become more salient, calling for robust frameworks to ensure responsible integration. This paper underscores the necessity for a balanced approach in implementing MLLMs, where the technology complements rather than supplants the educator's role, ensuring thus an effective and ethical use of AI in science education. It calls for further research to explore the nuanced implications of MLLMs on the evolving role of educators and to extend the discourse beyond science education to other disciplines. Through the exploration of potentials, challenges, and future implications, we aim to contribute to a preliminary understanding of the transformative trajectory of MLLMs in science education and beyond.

The emergence of ChatGPT and other large language models (LLMs) has greatly increased interest in utilizing LLMs as therapists to support individuals struggling with mental health challenges. However, due to the lack of systematic studies, our understanding of how LLM therapists behave, i.e., ways in which they respond to clients, is significantly limited. Understanding their behavior across a wide range of clients and situations is crucial to accurately assess their capabilities and limitations in the high-risk setting of mental health, where undesirable behaviors can lead to severe consequences. In this paper, we propose BOLT, a novel computational framework to study the conversational behavior of LLMs when employed as therapists. We develop an in-context learning method to quantitatively measure the behavior of LLMs based on 13 different psychotherapy techniques including reflections, questions, solutions, normalizing, and psychoeducation. Subsequently, we compare the behavior of LLM therapists against that of high- and low-quality human therapy, and study how their behavior can be modulated to better reflect behaviors observed in high-quality therapy. Our analysis of GPT and Llama-variants reveals that these LLMs often resemble behaviors more commonly exhibited in low-quality therapy rather than high-quality therapy, such as offering a higher degree of problem-solving advice when clients share emotions, which is against typical recommendations. At the same time, unlike low-quality therapy, LLMs reflect significantly more upon clients' needs and strengths. Our analysis framework suggests that despite the ability of LLMs to generate anecdotal examples that appear similar to human therapists, LLM therapists are currently not fully consistent with high-quality care, and thus require additional research to ensure quality care.

The high cost of full-parameter fine-tuning (FFT) of Large Language Models (LLMs) has led to a series of parameter-efficient fine-tuning (PEFT) methods. However, it remains unclear which methods provide the best cost-performance trade-off at different model scales. We introduce Astraios, a suite of 28 instruction-tuned OctoCoder models using 7 tuning methods and 4 model sizes up to 16 billion parameters. Through investigations across 5 tasks and 8 different datasets encompassing both code comprehension and code generation tasks, we find that FFT generally leads to the best downstream performance across all scales, and PEFT methods differ significantly in their efficacy based on the model scale. LoRA usually offers the most favorable trade-off between cost and performance. Further investigation into the effects of these methods on both model robustness and code security reveals that larger models tend to demonstrate reduced robustness and less security. At last, we explore the relationships among updated parameters, cross-entropy loss, and task performance. We find that the tuning effectiveness observed in small models generalizes well to larger models, and the validation loss in instruction tuning can be a reliable indicator of overall downstream performance.

Temporal validity is an important property of text that is useful for many downstream applications, such as recommender systems, conversational AI, or story understanding. Existing benchmarking tasks often require models to identify the temporal validity duration of a single statement. However, in many cases, additional contextual information, such as sentences in a story or posts on a social media profile, can be collected from the available text stream. This contextual information may greatly alter the duration for which a statement is expected to be valid. We propose Temporal Validity Change Prediction, a natural language processing task benchmarking the capability of machine learning models to detect contextual statements that induce such change. We create a dataset consisting of temporal target statements sourced from Twitter and crowdsource sample context statements. We then benchmark a set of transformer-based language models on our dataset. Finally, we experiment with temporal validity duration prediction as an auxiliary task to improve the performance of the state-of-the-art model.

Large Language Models (LLMs) like ChatGPT are foundational in various applications due to their extensive knowledge from pre-training and fine-tuning. Despite this, they are prone to generating factual and commonsense errors, raising concerns in critical areas like healthcare, journalism, and education to mislead users. Current methods for evaluating LLMs' veracity are limited by test data leakage or the need for extensive human labor, hindering efficient and accurate error detection. To tackle this problem, we introduce a novel, automatic testing framework, FactChecker, aimed at uncovering factual inaccuracies in LLMs. This framework involves three main steps: First, it constructs a factual knowledge graph by retrieving fact triplets from a large-scale knowledge database. Then, leveraging the knowledge graph, FactChecker employs a rule-based approach to generates three types of questions (Yes-No, Multiple-Choice, and WH questions) that involve single-hop and multi-hop relations, along with correct answers. Lastly, it assesses the LLMs' responses for accuracy using tailored matching strategies for each question type. Our extensive tests on six prominent LLMs, including text-davinci-002, text-davinci-003, ChatGPT~(gpt-3.5-turbo, gpt-4), Vicuna, and LLaMA-2, reveal that FactChecker can trigger factual errors in up to 45\% of questions in these models. Moreover, we demonstrate that FactChecker's test cases can improve LLMs' factual accuracy through in-context learning and fine-tuning (e.g., llama-2-13b-chat's accuracy increase from 35.3\% to 68.5\%). We are making all code, data, and results available for future research endeavors.

Recent advancements in large language models (LLMs) have propelled Artificial Intelligence (AI) to new heights, enabling breakthroughs in various tasks such as writing assistance, code generation, and machine translation. A significant distinction of advanced LLMs, such as ChatGPT, is their demonstrated ability to "reason." However, evaluating the reasoning ability of LLMs remains a challenge as most existing evaluations focus on their accuracy on the downstream tasks rather than directly assessing their reasoning processes. Efforts have been made to develop benchmarks and metrics to assess reasoning in LLMs, but they suffer from data leakage or limited scope. In this paper, we introduce LogicAsker, an automatic approach that comprehensively evaluates and improves the logical reasoning abilities of LLMs under a set of atomic reasoning skills based on propositional and predicate logic. The results provide insights into LLMs' reasoning abilities and reveal the logical rules the LLMs did not learn well. We evaluate LogicAsker on six widely deployed LLMs, including GPT-3, ChatGPT, GPT-4, Bard, Vicuna, and Guanaco. The results show that test cases from LogicAsker can find logical reasoning failures in different LLMs with a rate of 25\% - 94\%. In addition, the test cases of LogicAsker can be further used to design demonstration examples for in-context learning, which effectively improves the logical reasoning ability of LLMs, e.g., 10\% for GPT-4. As far as we know, our work is the first to create prompts based on testing results to improve LLMs' formal reasoning ability effectively. All the code, data, and results will be released for reproduction and future research.

String matching algorithms in the presence of abbreviations, such as in Stock Keeping Unit (SKU) product catalogs, remains a relatively unexplored topic. In this paper, we present a unified architecture for SKU search that provides both a real-time suggestion system (based on a Trie data structure) as well as a lower latency search system (making use of character level TF-IDF in combination with language model vector embeddings) where users initiate the search process explicitly. We carry out ablation studies that justify designing a complex search system composed of multiple components to address the delicate trade-off between speed and accuracy. Using SKU search in the Dynamics CRM as an example, we show how our system vastly outperforms, in all aspects, the results provided by the default search engine. Finally, we show how SKU descriptions may be enhanced via generative text models (using gpt-3.5-turbo)

This paper describes the architecture and systems built towards solving the SemEval 2023 Task 2: MultiCoNER II (Multilingual Complex Named Entity Recognition) [1]. We evaluate two approaches (a) a traditional Conditional Random Fields model and (b) a Large Language Model (LLM) fine-tuned with a customized head and compare the two approaches. The novel ideas explored are: 1) Decaying auxiliary loss (with residual) - where we train the model on an auxiliary task of Coarse-Grained NER and include this task as a part of the loss function 2) Triplet token blending - where we explore ways of blending the embeddings of neighboring tokens in the final NER layer prior to prediction 3) Task-optimal heads - where we explore a variety of custom heads and learning rates for the final layer of the LLM. We also explore multiple LLMs including GPT-3 and experiment with a variety of dropout and other hyperparameter settings before arriving at our final model which achieves micro & macro f1 of 0.85/0.84 (on dev) and 0.67/0.61 on the test data . We show that while pre-trained LLMs, by themselves, bring about a large improvement in scores as compared to traditional models, we also demonstrate that tangible improvements to the Macro-F1 score can be made by augmenting the LLM with additional feature/loss/model engineering techniques described above.