Large language models (LLMs), such as ChatGPT and GPT-4, are gaining wide-spread real world use. Yet, these LLMs are closed source, and little is known about their performance in real-world use cases. In this paper, we apply and evaluate the combination of ChatGPT and GPT-4 for the real-world task of mining insights from a text corpus in order to identify research challenges in the field of HCI. We extract 4,392 research challenges in over 100 topics from the 2023 CHI conference proceedings and visualize the research challenges for interactive exploration. We critically evaluate the LLMs on this practical task and conclude that the combination of ChatGPT and GPT-4 makes an excellent cost-efficient means for analyzing a text corpus at scale. Cost-efficiency is key for flexibly prototyping research ideas and analyzing text corpora from different perspectives, with implications for applying LLMs for mining insights in academia and practice.

First, Melissa tells me, there is a lot of disagreement about foundation models, which has taken up most of the energy and space during the latest debates. There are several definitions of the term "foundation model" floating around, which is part of what's causing the discord, but the core concept has to do with general-purpose AI that can do many different things for various applications. You've probably played around with ChatGPT; that interface is essentially powered by a foundation model, in this case a large language model from OpenAI. Making this more complex, though, is that these technologies can also be plugged into various other applications with more narrow uses, like education or advertising. Initial versions of the EU AI Act didn't explicitly consider foundation models, but Melissa notes that the proliferation of generative AI products over the past year pushed lawmakers to integrate them into the risk framework. In the version of the legislation passed by Parliament in June, all foundation models would be tightly regulated regardless of their assigned risk category or how they are used. This was deemed necessary in light of the vast amount of training data required to build them, as well as IP and privacy concerns and the overall impact they have on other technologies. But of course, tech companies that build foundation models have disputed this and advocate for a more nuanced approach that considers how the models are used. France, Germany, and Italy have flipped their positions and gone so far to say that foundation models should be largely exempt from AI Act regulations.

We design a user-friendly and scalable knowledge graph construction (KGC) system for extracting structured knowledge from the unstructured corpus. Different from existing KGC systems, gBuilder provides a flexible and user-defined pipeline to embrace the rapid development of IE models. More built-in template-based or heuristic operators and programmable operators are available for adapting to data from different domains. Furthermore, we also design a cloud-based self-adaptive task scheduling for gBuilder to ensure its scalability on large-scale knowledge graph construction. Experimental evaluation demonstrates the ability of gBuilder to organize multiple information extraction models for knowledge graph construction in a uniform platform, and confirms its high scalability on large-scale KGC tasks.

Unit tests play a key role in ensuring the correctness of software. However, manually creating unit tests is a laborious task, motivating the need for automation. Large Language Models (LLMs) have recently been applied to this problem, utilizing additional training or few-shot learning on examples of existing tests. This paper presents a large-scale empirical evaluation on the effectiveness of LLMs for automated unit test generation without additional training or manual effort, providing the LLM with the signature and implementation of the function under test, along with usage examples extracted from documentation. We also attempt to repair failed generated tests by re-prompting the model with the failing test and error message. We implement our approach in TestPilot, a test generation tool for JavaScript that automatically generates unit tests for all API functions in an npm package. We evaluate TestPilot using OpenAI's gpt3.5-turbo LLM on 25 npm packages with a total of 1,684 API functions. The generated tests achieve a median statement coverage of 70.2% and branch coverage of 52.8%, significantly improving on Nessie, a recent feedback-directed JavaScript test generation technique, which achieves only 51.3% statement coverage and 25.6% branch coverage. We also find that 92.8% of TestPilot's generated tests have no more than 50% similarity with existing tests (as measured by normalized edit distance), with none of them being exact copies. Finally, we run TestPilot with two additional LLMs, OpenAI's older code-cushman-002 LLM and the open LLM StarCoder. Overall, we observed similar results with the former (68.2% median statement coverage), and somewhat worse results with the latter (54.0% median statement coverage), suggesting that the effectiveness of the approach is influenced by the size and training set of the LLM, but does not fundamentally depend on the specific model.

Caution: This paper includes offensive words that could potentially cause unpleasantness. The fast-paced evolution of generative language models such as GPT-4 has demonstrated outstanding results in various NLP generation tasks. However, due to the potential generation of offensive words related to race or gender, various Controllable Text Generation (CTG) methods have been proposed to mitigate the occurrence of harmful words. However, existing CTG methods not only reduce toxicity but also negatively impact several aspects of the language model's generation performance, including topic consistency, grammar, and perplexity. This paper explores the limitations of previous methods and introduces a novel solution in the form of a simple Gated Toxicity Avoidance (GTA) that can be applied to any CTG method. We also evaluate the effectiveness of the proposed GTA by comparing it with state-of-the-art CTG methods across various datasets. Our findings reveal that gated toxicity avoidance efficiently achieves comparable levels of toxicity reduction to the original CTG methods while preserving the generation performance of the language model.

Dense retrieval has become a prominent method to obtain relevant context or world knowledge in open-domain NLP tasks. When we use a learned dense retriever on a retrieval corpus at inference time, an often-overlooked design choice is the retrieval unit in which the corpus is indexed, e.g. document, passage, or sentence. We discover that the retrieval unit choice significantly impacts the performance of both retrieval and downstream tasks. Distinct from the typical approach of using passages or sentences, we introduce a novel retrieval unit, proposition, for dense retrieval. Propositions are defined as atomic expressions within text, each encapsulating a distinct factoid and presented in a concise, self-contained natural language format. We conduct an empirical comparison of different retrieval granularity. Our results reveal that proposition-based retrieval significantly outperforms traditional passage or sentence-based methods in dense retrieval. Moreover, retrieval by proposition also enhances the performance of downstream QA tasks, since the retrieved texts are more condensed with question-relevant information, reducing the need for lengthy input tokens and minimizing the inclusion of extraneous, irrelevant information.

We explore the abstract reasoning abilities of text-only and multimodal versions of GPT-4, using the ConceptARC benchmark [10], which is designed to evaluate robust understanding and reasoning with core-knowledge concepts. We extend the work of Moskvichev et al. [10] by evaluating GPT-4 on more detailed, one-shot prompting (rather than simple, zero-shot prompts) with text versions of ConceptARC tasks, and by evaluating GPT-4V, the multimodal version of GPT-4, on zero- and one-shot prompts using image versions of the simplest tasks. Our experimental results support the conclusion that neither version of GPT-4 has developed robust abstraction abilities at humanlike levels.

Recent advancements in large language models (LLMs) have spurred interest in using them for generating robot programs from natural language, with promising initial results. We investigate the use of LLMs to generate programs for service mobile robots leveraging mobility, perception, and human interaction skills, and where accurate sequencing and ordering of actions is crucial for success. We contribute CodeBotler, an open-source robot-agnostic tool to program service mobile robots from natural language, and RoboEval, a benchmark for evaluating LLMs' capabilities of generating programs to complete service robot tasks. CodeBotler performs program generation via few-shot prompting of LLMs with an embedded domain-specific language (eDSL) in Python, and leverages skill abstractions to deploy generated programs on any general-purpose mobile robot. RoboEval evaluates the correctness of generated programs by checking execution traces starting with multiple initial states, and checking whether the traces satisfy temporal logic properties that encode correctness for each task. RoboEval also includes multiple prompts per task to test for the robustness of program generation. We evaluate several popular state-of-the-art LLMs with the RoboEval benchmark, and perform a thorough analysis of the modes of failures, resulting in a taxonomy that highlights common pitfalls of LLMs at generating robot programs. We release our code and benchmark at https://amrl.cs.utexas.edu/codebotler/.

Instruction tuning large language models (LLMs) using machine-generated instruction-following data has been shown to improve zero-shot capabilities on new tasks, but the idea is less explored in the multimodal field. We present the first attempt to use language-only GPT-4 to generate multimodal language-image instruction-following data. By instruction tuning on such generated data, we introduce LLaVA: Large Language and Vision Assistant, an end-to-end trained large multimodal model that connects a vision encoder and an LLM for generalpurpose visual and language understanding. To facilitate future research on visual instruction following, we construct two evaluation benchmarks with diverse and challenging application-oriented tasks. Our experiments show that LLaVA demonstrates impressive multimodal chat abilities, sometimes exhibiting the behaviors of multimodal GPT-4 on unseen images/instructions, and yields a 85.1% relative score compared with GPT-4 on a synthetic multimodal instruction-following dataset. When fine-tuned on Science QA, the synergy of LLaVA and GPT-4 achieves a new state-of-the-art accuracy of 92.53%. We make GPT-4 generated visual instruction tuning data, our model, and code publicly available.

We introduce a method to train vision-language models for remote-sensing images without using any textual annotations. Our key insight is to use co-located internet imagery taken on the ground as an intermediary for connecting remote-sensing images and language. Specifically, we train an image encoder for remote sensing images to align with the image encoder of CLIP using a large amount of paired internet and satellite images. Our unsupervised approach enables the training of a first-of-its-kind large-scale vision language model (VLM) for remote sensing images at two different resolutions. We show that these VLMs enable zero-shot, open-vocabulary image classification, retrieval, segmentation and visual question answering for satellite images. On each of these tasks, our VLM trained without textual annotations outperforms existing VLMs trained with supervision, with gains of up to 20% for classification and 80% for segmentation. Our planet is constantly captured by an extensive array of remote sensors such as satellites or drones. These earth observation images enable the monitoring of various events on the earth such as deforestation, forest fires, and droughts so that rapid actions can be taken to protect our environment. While these images can shed light on various insights about our planet, the scale of such data is huge. This has prompted the development of automatic analysis models that could extract relevant information from a large amount of remotely sensed images. While useful, these models are often specialized and can only recognize a pre-defined set of concepts. Besides, they could be complex, decreasing their accessibility to experts outside of the domain of artificial intelligence. Researchers developing automatic analysis methods for internet imagery encountered a similar problem a few years ago. One promising solution is to leverage large-scale vision-language models (VLMs) that are trained on millions or even billions of text-image pairs collected on the internet (Radford et al., 2021; Li et al., 2023). These models have demonstrated remarkable abilities to perform open-vocabulary recognition (Gu et al., 2022; Kuo et al., 2023) and enhance accessibility to non-AI experts (Alayrac et al., 2022; Surís et al., 2023). It would be incredibly valuable for a range of applications to replicate the success of openvocabulary recognition for satellite images as well, allowing an analyst to simply query, say, "Where are all the farmlands in the state of Massachusetts?" without requiring any new training or annotation for farms.