ChatGPT embarks on a new era of artificial intelligence and will revolutionize the way we approach intelligent traffic safety systems. This paper begins with a brief introduction about the development of large language models (LLMs). Next, we exemplify using ChatGPT to address key traffic safety issues. Furthermore, we discuss the controversies surrounding LLMs, raise critical questions for their deployment, and provide our solutions. Moreover, we propose an idea of multi-modality representation learning for smarter traffic safety decision-making and open more questions for application improvement. We believe that LLM will both shape and potentially facilitate components of traffic safety research.
Large Language Models (LLMs) have sparked significant interest in their generative capabilities, leading to the development of various commercial applications. The high cost of using the models drives application builders to maximize the value of generation under a limited inference budget. This paper presents a study of optimizing inference hyperparameters such as the number of responses, temperature and max tokens, which significantly affects the utility/cost of text generation. We design a framework named EcoOptiGen which leverages economical hyperparameter optimization and cost-based pruning. Experiments with the GPT-3.5/GPT-4 models on a variety of tasks verify its effectiveness. EcoOptiGen is implemented in the `autogen' package of the FLAML library: \url{https://aka.ms/autogen}.
Neural text-to-speech (TTS) generally consists of cascaded architecture with separately optimized acoustic model and vocoder, or end-to-end architecture with continuous mel-spectrograms or self-extracted speech frames as the intermediate representations to bridge acoustic model and vocoder, which suffers from two limitations: 1) the continuous acoustic frames are hard to predict with phoneme only, and acoustic information like duration or pitch is also needed to solve the one-to-many problem, which is not easy to scale on large scale and noise datasets; 2) to achieve diverse speech output based on continuous speech features, complex VAE or flow-based models are usually required. In this paper, we propose FoundationTTS, a new speech synthesis system with a neural audio codec for discrete speech token extraction and waveform reconstruction and a large language model for discrete token generation from linguistic (phoneme) tokens. Specifically, 1) we propose a hierarchical codec network based on vector-quantized auto-encoders with adversarial training (VQ-GAN), which first extracts continuous frame-level speech representations with fine-grained codec, and extracts a discrete token from each continuous speech frame with coarse-grained codec; 2) we jointly optimize speech token, linguistic tokens, speaker token together with a large language model and predict the discrete speech tokens autoregressively. Experiments show that FoundationTTS achieves a MOS gain of +0.14 compared to the baseline system. In ASR customization tasks, our method achieves 7.09\% and 10.35\% WERR respectively over two strong customized ASR baselines.
In recent years, In-context Learning (ICL) has gained increasing attention and emerged as the new paradigm for large language model (LLM) evaluation. Unlike traditional fine-tuning methods, ICL instead adapts the pre-trained models to unseen tasks without any parameter updates. However, the implementation of ICL is sophisticated due to the diverse retrieval and inference methods involved, as well as the varying pre-processing requirements for different models, datasets, and tasks. A unified and flexible framework for ICL is urgently needed to ease the implementation of the aforementioned components. To facilitate ICL research, we introduce OpenICL, an open-source toolkit for ICL and LLM evaluation. OpenICL is research-friendly with a highly flexible architecture that users can easily combine different components to suit their needs. It also provides various state-of-the-art retrieval and inference methods to streamline the process of adapting ICL to cutting-edge research. The effectiveness of OpenICL has been validated on a wide range of NLP tasks, including classification, QA, machine translation, and semantic parsing. As a side-product, we found OpenICL to be an efficient yet robust tool for LLMs evaluation. OpenICL is released at https://github.com/Shark-NLP/OpenICL
Large language models excel at a wide range of complex tasks. However, enabling general inference in the real world, e.g., for robotics problems, raises the challenge of grounding. We propose embodied language models to directly incorporate real-world continuous sensor modalities into language models and thereby establish the link between words and percepts. Input to our embodied language model are multi-modal sentences that interleave visual, continuous state estimation, and textual input encodings. We train these encodings end-to-end, in conjunction with a pre-trained large language model, for multiple embodied tasks including sequential robotic manipulation planning, visual question answering, and captioning. Our evaluations show that PaLM-E, a single large embodied multimodal model, can address a variety of embodied reasoning tasks, from a variety of observation modalities, on multiple embodiments, and further, exhibits positive transfer: the model benefits from diverse joint training across internet-scale language, vision, and visual-language domains. Our largest model, PaLM-E-562B with 562B parameters, in addition to being trained on robotics tasks, is a visual-language generalist with state-of-the-art performance on OK-VQA, and retains generalist language capabilities with increasing scale.
Knowledge-based visual question answering (VQA) requires external knowledge beyond the image to answer the question. Early studies retrieve required knowledge from explicit knowledge bases (KBs), which often introduces irrelevant information to the question, hence restricting the performance of their models. Recent works have resorted to using a powerful large language model (LLM) as an implicit knowledge engine to acquire the necessary knowledge for answering. Despite the encouraging results achieved by these methods, we argue that they have not fully activated the capacity of the \emph{blind} LLM as the provided textual input is insufficient to depict the required visual information to answer the question. In this paper, we present Prophet -- a conceptually simple, flexible, and general framework designed to prompt LLM with answer heuristics for knowledge-based VQA. Specifically, we first train a vanilla VQA model on a specific knowledge-based VQA dataset without external knowledge. After that, we extract two types of complementary answer heuristics from the VQA model: answer candidates and answer-aware examples. The two types of answer heuristics are jointly encoded into a formatted prompt to facilitate the LLM's understanding of both the image and question, thus generating a more accurate answer. By incorporating the state-of-the-art LLM GPT-3, Prophet significantly outperforms existing state-of-the-art methods on four challenging knowledge-based VQA datasets. Prophet is general that can be instantiated with the combinations of different VQA models (i.e., both discriminative and generative ones) and different LLMs (i.e., both commercial and open-source ones). Moreover, Prophet can also be integrated with modern large multimodal models in different stages, which is named Prophet++, to further improve the capabilities on knowledge-based VQA tasks.
ChatGPT is a groundbreaking ``chatbot"--an AI interface built on a large language model that was trained on an enormous corpus of human text to emulate human conversation. Beyond its ability to converse in a plausible way, it has attracted attention for its ability to competently answer questions from the bar exam and from MBA coursework, and to provide useful assistance in writing computer code. These apparent abilities have prompted discussion of ChatGPT as both a threat to the integrity of higher education and conversely as a powerful teaching tool. In this work we present a preliminary analysis of how two versions of ChatGPT (ChatGPT3.5 and ChatGPT4) fare in the field of first-semester university physics, using a modified version of the Force Concept Inventory (FCI) to assess whether it can give correct responses to conceptual physics questions about kinematics and Newtonian dynamics. We demonstrate that, by some measures, ChatGPT3.5 can match or exceed the median performance of a university student who has completed one semester of college physics, though its performance is notably uneven and the results are nuanced. By these same measures, we find that ChatGPT4's performance is approaching the point of being indistinguishable from that of an expert physicist when it comes to introductory mechanics topics. After the completion of our work we became aware of Ref [1], which preceded us to publication and which completes an extensive analysis of the abilities of ChatGPT3.5 in a physics class, including a different modified version of the FCI. We view this work as confirming that portion of their results, and extending the analysis to ChatGPT4, which shows rapid and notable improvement in most, but not all respects.
Large-language models have recently demonstrated impressive zero-shot capabilities in a variety of natural language tasks such as summarization, dialogue generation, and question-answering. Despite many promising applications in clinical medicine, adoption of these models in real-world settings has been largely limited by their tendency to generate incorrect and sometimes even toxic statements. In this study, we develop Almanac, a large language model framework augmented with retrieval capabilities for medical guideline and treatment recommendations. Performance on a novel dataset of clinical scenarios (n = 130) evaluated by a panel of 5 board-certified and resident physicians demonstrates significant increases in factuality (mean of 18% at p-value < 0.05) across all specialties, with improvements in completeness and safety. Our results demonstrate the potential for large language models to be effective tools in the clinical decision-making process, while also emphasizing the importance of careful testing and deployment to mitigate their shortcomings.
Reward design in reinforcement learning (RL) is challenging since specifying human notions of desired behavior may be difficult via reward functions or require many expert demonstrations. Can we instead cheaply design rewards using a natural language interface? This paper explores how to simplify reward design by prompting a large language model (LLM) such as GPT-3 as a proxy reward function, where the user provides a textual prompt containing a few examples (few-shot) or a description (zero-shot) of the desired behavior. Our approach leverages this proxy reward function in an RL framework. Specifically, users specify a prompt once at the beginning of training. During training, the LLM evaluates an RL agent's behavior against the desired behavior described by the prompt and outputs a corresponding reward signal. The RL agent then uses this reward to update its behavior. We evaluate whether our approach can train agents aligned with user objectives in the Ultimatum Game, matrix games, and the DealOrNoDeal negotiation task. In all three tasks, we show that RL agents trained with our framework are well-aligned with the user's objectives and outperform RL agents trained with reward functions learned via supervised learning
We study the performance of a commercially available large language model (LLM) known as ChatGPT on math word problems (MWPs) from the dataset DRAW-1K. To our knowledge, this is the first independent evaluation of ChatGPT. We found that ChatGPT's performance changes dramatically based on the requirement to show its work, failing 20% of the time when it provides work compared with 84% when it does not. Further several factors about MWPs relating to the number of unknowns and number of operations that lead to a higher probability of failure when compared with the prior, specifically noting (across all experiments) that the probability of failure increases linearly with the number of addition and subtraction operations. We also have released the dataset of ChatGPT's responses to the MWPs to support further work on the characterization of LLM performance and present baseline machine learning models to predict if ChatGPT can correctly answer an MWP. We have released a dataset comprised of ChatGPT's responses to support further research in this area.
Programming skills are rapidly becoming essential for many educational paths and career opportunities. Yet, for many international students, the traditional approach to teaching introductory programming courses can be a significant challenge due to the complexities of the language, the lack of prior programming knowledge, and the language and cultural barriers. This study explores how large language models and gamification can scaffold coding learning and increase Chinese students sense of belonging in introductory programming courses. In this project, a gamification intelligent tutoring system was developed to adapt to Chinese international students learning needs and provides scaffolding to support their success in introductory computer programming courses.
This paper presents a novel optimization framework for automatic speech recognition (ASR) with the aim of reducing hallucinations produced by an ASR model. The proposed framework optimizes the ASR model to maximize an expected factual consistency score between ASR hypotheses and ground-truth transcriptions, where the factual consistency score is computed by a separately trained estimator. Experimental results using the AMI meeting corpus and the VoxPopuli corpus show that the ASR model trained with the proposed framework generates ASR hypotheses that have significantly higher consistency scores with ground-truth transcriptions while maintaining the word error rates close to those of cross entropy-trained ASR models. Furthermore, it is shown that training the ASR models with the proposed framework improves the speech summarization quality as measured by the factual consistency of meeting conversation summaries generated by a large language model.
Deep neural networks drive the success of natural language processing. A fundamental property of language is its compositional structure, allowing humans to systematically produce forms for new meanings. For humans, languages with more compositional and transparent structures are typically easier to learn than those with opaque and irregular structures. However, this learnability advantage has not yet been shown for deep neural networks, limiting their use as models for human language learning. Here, we directly test how neural networks compare to humans in learning and generalizing different languages that vary in their degree of compositional structure. We evaluate the memorization and generalization capabilities of a large language model and recurrent neural networks, and show that both deep neural networks exhibit a learnability advantage for more structured linguistic input: neural networks exposed to more compositional languages show more systematic generalization, greater agreement between different agents, and greater similarity to human learners.
In Federated Learning (FL), accessing private client data incurs communication and privacy costs. As a result, FL deployments commonly prefinetune pretrained foundation models on a (large, possibly public) dataset that is held by the central server; they then FL-finetune the model on a private, federated dataset held by clients. Evaluating prefinetuning dataset quality reliably and privately is therefore of high importance. To this end, we propose FreD (Federated Private Fr\'echet Distance) -- a privately computed distance between a prefinetuning dataset and federated datasets. Intuitively, it privately computes and compares a Fr\'echet distance between embeddings generated by a large language model on both the central (public) dataset and the federated private client data. To make this computation privacy-preserving, we use distributed, differentially-private mean and covariance estimators. We show empirically that FreD accurately predicts the best prefinetuning dataset at minimal privacy cost. Altogether, using FreD we demonstrate a proof-of-concept for a new approach in private FL training: (1) customize a prefinetuning dataset to better match user data (2) prefinetune (3) perform FL-finetuning.
A rapidly increasing amount of human conversation occurs online. But divisiveness and conflict can fester in text-based interactions on social media platforms, in messaging apps, and on other digital forums. Such toxicity increases polarization and, importantly, corrodes the capacity of diverse societies to develop efficient solutions to complex social problems that impact everyone. Scholars and civil society groups promote interventions that can make interpersonal conversations less divisive or more productive in offline settings, but scaling these efforts to the amount of discourse that occurs online is extremely challenging. We present results of a large-scale experiment that demonstrates how online conversations about divisive topics can be improved with artificial intelligence tools. Specifically, we employ a large language model to make real-time, evidence-based recommendations intended to improve participants' perception of feeling understood in conversations. We find that these interventions improve the reported quality of the conversation, reduce political divisiveness, and improve the tone, without systematically changing the content of the conversation or moving people's policy attitudes. These findings have important implications for future research on social media, political deliberation, and the growing community of scholars interested in the place of artificial intelligence within computational social science.
We test whether Large Language Models (LLMs) can be used to simulate human participants in social-science studies. To do this, we run replications of 14 studies from the Many Labs 2 replication project with OpenAI's text-davinci-003 model, colloquially known as GPT3.5. Based on our pre-registered analyses, we find that among the eight studies we could analyse, our GPT sample replicated 37.5% of the original results and 37.5% of the Many Labs 2 results. However, we were unable to analyse the remaining six studies due to an unexpected phenomenon we call the "correct answer" effect. Different runs of GPT3.5 answered nuanced questions probing political orientation, economic preference, judgement, and moral philosophy with zero or near-zero variation in responses: with the supposedly "correct answer." In one exploratory follow-up study, we found that a "correct answer" was robust to changing the demographic details that precede the prompt. In another, we found that most but not all "correct answers" were robust to changing the order of answer choices. One of our most striking findings occurred in our replication of the Moral Foundations Theory survey results, where we found GPT3.5 identifying as a political conservative in 99.6% of the cases, and as a liberal in 99.3% of the cases in the reverse-order condition. However, both self-reported 'GPT conservatives' and 'GPT liberals' showed right-leaning moral foundations. Our results cast doubts on the validity of using LLMs as a general replacement for human participants in the social sciences. Our results also raise concerns that a hypothetical AI-led future may be subject to a diminished diversity-of-thought.
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.
Recent advancements in large scale text-to-image models have opened new possibilities for guiding the creation of images through human-devised natural language. However, while prior literature has primarily focused on the generation of individual images, it is essential to consider the capability of these models to ensure coherency within a sequence of images to fulfill the demands of real-world applications such as storytelling. To address this, here we present a novel neural pipeline for generating a coherent storybook from the plain text of a story. Specifically, we leverage a combination of a pre-trained Large Language Model and a text-guided Latent Diffusion Model to generate coherent images. While previous story synthesis frameworks typically require a large-scale text-to-image model trained on expensive image-caption pairs to maintain the coherency, we employ simple textual inversion techniques along with detector-based semantic image editing which allows zero-shot generation of the coherent storybook. Experimental results show that our proposed method outperforms state-of-the-art image editing baselines.
A previous paper presented Witscript, a system for generating conversational jokes that rely on wordplay. This paper extends that work by presenting Witscript 2, which uses a large language model to generate conversational jokes that rely on common sense instead of wordplay. Like Witscript, Witscript 2 is based on joke-writing algorithms created by an expert comedy writer. Human evaluators judged Witscript 2's responses to input sentences to be jokes 46% of the time, compared to 70% of the time for human-written responses. This is evidence that Witscript 2 represents another step toward giving a chatbot a humanlike sense of humor.
Can large language models be trained to produce philosophical texts that are difficult to distinguish from texts produced by human philosophers? To address this question, we fine-tuned OpenAI's GPT-3 with the works of philosopher Daniel C. Dennett as additional training data. To explore the Dennett model, we asked the real Dennett ten philosophical questions and then posed the same questions to the language model, collecting four responses for each question without cherry-picking. We recruited 425 participants to distinguish Dennett's answer from the four machine-generated answers. Experts on Dennett's work (N = 25) succeeded 51% of the time, above the chance rate of 20% but short of our hypothesized rate of 80% correct. For two of the ten questions, the language model produced at least one answer that experts selected more frequently than Dennett's own answer. Philosophy blog readers (N = 302) performed similarly to the experts, while ordinary research participants (N = 98) were near chance distinguishing GPT-3's responses from those of an "actual human philosopher".