Large Language models (LLMs) can be induced to solve non-trivial problems with "few-shot" prompts including illustrative problem-solution examples. Now if the few-shots also include "chain of thought" (CoT) explanations, which are of the form problem-explanation-solution, LLMs will generate a "explained" solution, and perform even better. Recently an exciting, substantially better technique, self-consistency [1] (S-C) has emerged, based on the intuition that there are many plausible explanations for the right solution; when the LLM is sampled repeatedly to generate a pool of explanation-solution pairs, for a given problem, the most frequently occurring solutions in the pool (ignoring the explanations) tend to be even more likely to be correct! Unfortunately, the use of this highly-performant S-C (or even CoT) approach in software engineering settings is hampered by the lack of explanations; most software datasets lack explanations. In this paper, we describe an application of the S-C approach to program repair, using the commit log on the fix as the explanation, only in the illustrative few-shots. We achieve state-of-the art results, beating previous approaches to prompting-based program repair, on the MODIT dataset; we also find evidence suggesting that the correct commit messages are helping the LLM learn to produce better patches.

Language models have been shown to perform better with an increase in scale on a wide variety of tasks via the in-context learning paradigm. In this paper, we investigate the hypothesis that the ability of a large language model to in-context learn-perform a task is not uniformly spread across all of its underlying components. Using a 66 billion parameter language model (OPT-66B) across a diverse set of 14 downstream tasks, we find this is indeed the case: $\sim$70% of attention heads and $\sim$20% of feed forward networks can be removed with minimal decline in task performance. We find substantial overlap in the set of attention heads (un)important for in-context learning across tasks and number of in-context examples. We also address our hypothesis through a task-agnostic lens, finding that a small set of attention heads in OPT-66B score highly on their ability to perform primitive induction operations associated with in-context learning, namely, prefix matching and copying. These induction heads overlap with task-specific important heads, reinforcing arguments by Olsson et al. (arXiv:2209.11895) regarding induction head generality to more sophisticated behaviors associated with in-context learning. Overall, our study provides several insights that indicate large language models may be under-trained for in-context learning and opens up questions on how to pre-train language models to more effectively perform in-context learning.

Multilingual large language models (LLMs) seem to work better in English. These AIs are designed to respond to queries in multiple languages but they respond better if asked to translate the request into English first. LLMs have become a key part of the artificial intelligence revolution since the release of ChatGPT by OpenAI in November 2022.

Content moderation has been one of the thorniest issues on the internet for decades. It's a difficult subject matter for anyone to tackle, considering the subjectivity that goes hand-in-hand with figuring out what content should be permissible on a given platform. ChatGPT maker OpenAI thinks it can help and it has been putting GPT-4's content moderation skills to the test. It's using the large multimodal model "to build a content moderation system that is scalable, consistent and customizable." The company wrote in a blog post that GPT-4 can not only help make content moderation decisions, but aid in developing policies and swiftly iterating on policy changes, "reducing the cycle from months to hours."

Like a lot of people, Alex Woolner was instantly fascinated by ChatGPT, the AI-powered chatbot which took the internet by storm when it launched in November 2022. But instead of asking it to write screenplays or satirical poetry or magazine articles (OK, not really), she turned it toward her hobby: making cute crochet animals called amigurumi. "Crochet patterns in and of themselves are kind of like a code," says Woolner, who works for a study-abroad organization and also cofounded an arts and activism collective called Attack Bear Press. "I wanted to see what ChatGPT would produce if I asked it for a crochet pattern." The results were … well … adjectives such as "monstrous" spring to mind.

Beijing is poised to implement sweeping new regulations for artificial intelligence services this week, trying to balance state control of the technology with enough support that its companies can become viable global competitors. The government issued 24 guidelines that require platform providers to register their services and conduct a security review before they're brought to market. Seven agencies will take responsibility for oversight, including the Cyberspace Administration of China and the National Development and Reform Commission. The final regulations are less onerous than an original draft from April, but they show China, like Europe, moving ahead with government oversight of what may be the most promising -- and controversial -- technology of the last 30 years. The U.S., by contrast, has no legislation under serious consideration even after industry leaders warned that AI poses a "risk of extinction" and OpenAI's Sam Altman urged Congress in public hearings to get involved.

We study the task of zero-shot vision-and-language navigation (ZS-VLN), a practical yet challenging problem in which an agent learns to navigate following a path described by language instructions without requiring any path-instruction annotation data. Normally, the instructions have complex grammatical structures and often contain various action descriptions (e.g., "proceed beyond", "depart from"). How to correctly understand and execute these action demands is a critical problem, and the absence of annotated data makes it even more challenging. Note that a well-educated human being can easily understand path instructions without the need for any special training. In this paper, we propose an action-aware zero-shot VLN method ($A^2$Nav) by exploiting the vision-and-language ability of foundation models. Specifically, the proposed method consists of an instruction parser and an action-aware navigation policy. The instruction parser utilizes the advanced reasoning ability of large language models (e.g., GPT-3) to decompose complex navigation instructions into a sequence of action-specific object navigation sub-tasks. Each sub-task requires the agent to localize the object and navigate to a specific goal position according to the associated action demand. To accomplish these sub-tasks, an action-aware navigation policy is learned from freely collected action-specific datasets that reveal distinct characteristics of each action demand. We use the learned navigation policy for executing sub-tasks sequentially to follow the navigation instruction. Extensive experiments show $A^2$Nav achieves promising ZS-VLN performance and even surpasses the supervised learning methods on R2R-Habitat and RxR-Habitat datasets.

In this paper, we investigate the in-context learning ability of retrieval-augmented encoder-decoder language models. We first conduct a comprehensive analysis of the state-of-the-art ATLAS model and identify its limitations in in-context learning, primarily due to a mismatch between pretraining and testing, as well as a restricted context length. To address these issues, we propose RAVEN, a model that combines retrieval-augmented masked language modeling and prefix language modeling. We further introduce Fusion-in-Context Learning to enhance the few-shot performance by enabling the model to leverage more in-context examples without requiring additional training or model modifications. Through extensive experiments, we demonstrate that RAVEN significantly outperforms ATLAS and achieves results comparable to the most advanced language models in certain scenarios, despite having substantially fewer parameters. Our work underscores the potential of retrieval-augmented encoder-decoder language models for in-context learning and encourages further research in this direction.

Recent progress in large language models (LLMs) like GPT-4 and PaLM-2 has brought significant advancements in addressing math reasoning problems. In particular, OpenAI's latest version of GPT-4, known as GPT-4 Code Interpreter, shows remarkable performance on challenging math datasets. In this paper, we explore the effect of code on enhancing LLMs' reasoning capability by introducing different constraints on the Code Usage Frequency of GPT-4 Code Interpreter. We found that its success can be largely attributed to its powerful skills in generating and executing code, evaluating the output of code execution, and rectifying its solution when receiving unreasonable outputs. Based on this insight, we propose a novel and effective prompting method, explicit code-based self-verification (CSV), to further boost the mathematical reasoning potential of GPT-4 Code Interpreter. This method employs a zero-shot prompt on GPT-4 Code Interpreter to encourage it to use code to self-verify its answers. In instances where the verification state registers as "False", the model shall automatically amend its solution, analogous to our approach of rectifying errors during a mathematics examination. Furthermore, we recognize that the states of the verification result indicate the confidence of a solution, which can improve the effectiveness of majority voting. With GPT-4 Code Interpreter and CSV, we achieve an impressive zero-shot accuracy on MATH dataset (53.9% 84.3%). Large language models (LLMs) (Brown et al., 2020; OpenAI, 2023; Anil et al., 2023) have shown impressive success in various tasks, such as common sense understanding and code generation. However, they still fall short in mathematical reasoning, often producing nonsensical or inaccurate content and struggling with complex calculations. Previous attempts to tackle these challenges include the Chain-of-Thought (CoT) (Wei et al., 2022) framework, which enhances LLMs' logical reasoning abilities by generating intermediate steps in their reasoning process.

Automated log analysis is crucial in modern software-intensive systems for ensuring reliability and resilience throughout software maintenance and engineering life cycles. Existing methods perform tasks such as log parsing and log anomaly detection by providing a single prediction value without interpretation. However, given the increasing volume of system events, the limited interpretability of analysis results hinders analysts' trust and their ability to take appropriate actions. Moreover, these methods require substantial in-domain training data, and their performance declines sharply (by up to 62.5%) in online scenarios involving unseen logs from new domains, a common occurrence due to rapid software updates. In this paper, we propose LogPrompt, a novel zero-shot and interpretable log analysis approach. LogPrompt employs large language models (LLMs) to perform zero-shot log analysis tasks via a suite of advanced prompt strategies tailored for log tasks, which enhances LLMs' performance by up to 107.5% compared with simple prompts. Experiments on nine publicly available evaluation datasets across two tasks demonstrate that LogPrompt, despite using no training data, outperforms existing approaches trained on thousands of logs by up to around 50%. We also conduct a human evaluation of LogPrompt's interpretability, with six practitioners possessing over 10 years of experience, who highly rated the generated content in terms of usefulness and readability (averagely 4.42/5). LogPrompt also exhibits remarkable compatibility with open-source and smaller-scale LLMs, making it flexible for practical deployment.