Large language models (LLMs) can achieve highly effective performance on various reasoning tasks by incorporating step-by-step chain-of-thought (CoT) prompting as demonstrations. However, the reasoning chains of demonstrations generated by LLMs are prone to errors, which can subsequently lead to incorrect reasoning during inference. Furthermore, inappropriate exemplars (overly simplistic or complex), can affect overall performance among varying levels of difficulty. We introduce Iter-CoT (Iterative bootstrapping in Chain-of-Thoughts Prompting), an iterative bootstrapping approach for selecting exemplars and generating reasoning chains. By utilizing iterative bootstrapping, our approach enables LLMs to autonomously rectify errors, resulting in more precise and comprehensive reasoning chains. Simultaneously, our approach selects challenging yet answerable questions accompanied by reasoning chains as exemplars with a moderate level of difficulty, which enhances the LLMs' generalizability across varying levels of difficulty. Experimental results indicate that Iter-CoT exhibits superiority, achieving competitive performance across three distinct reasoning tasks on ten datasets.

Large Language Models (LLMs) have revolutionized natural language processing and demonstrated impressive capabilities in various tasks. Unfortunately, they are prone to hallucinations, where the model exposes incorrect or false information in its responses, which renders diligent evaluation approaches mandatory. While LLM performance in specific knowledge fields is often evaluated based on question and answer (Q&A) datasets, such evaluations usually report only a single accuracy number for the dataset, which often covers an entire field. This field-based evaluation, is problematic with respect to transparency and model improvement. A stratified evaluation could instead reveal subfields, where hallucinations are more likely to occur and thus help to better assess LLMs' risks and guide their further development. To support such stratified evaluations, we propose LLMMaps as a novel visualization technique that enables users to evaluate LLMs' performance with respect to Q&A datasets. LLMMaps provide detailed insights into LLMs' knowledge capabilities in different subfields, by transforming Q&A datasets as well as LLM responses into an internal knowledge structure. An extension for comparative visualization furthermore, allows for the detailed comparison of multiple LLMs. To assess LLMMaps we use them to conduct a comparative analysis of several state-of-the-art LLMs, such as BLOOM, GPT-2, GPT-3, ChatGPT and LLaMa-13B, as well as two qualitative user evaluations. All necessary source code and data for generating LLMMaps to be used in scientific publications and elsewhere is available on GitHub: https://github.com/viscom-ulm/LLMMaps

The recent emergence of Large Language Models based on the Transformer architecture has enabled dramatic advancements in the field of Natural Language Processing. However, these models have long inference latency, which limits their deployment and makes them prohibitively expensive for various real-time applications. The inference latency is further exacerbated by autoregressive generative tasks, as models need to run iteratively to generate tokens sequentially without leveraging token-level parallelization. To address this, we propose Big Little Decoder (BiLD), a framework that can improve inference efficiency and latency for a wide range of text generation applications. The BiLD framework contains two models with different sizes that collaboratively generate text. The small model runs autoregressively to generate text with a low inference cost, and the large model is only invoked occasionally to refine the small model's inaccurate predictions in a non-autoregressive manner. To coordinate the small and large models, BiLD introduces two simple yet effective policies: (1) the fallback policy that determines when to hand control over to the large model; and (2) the rollback policy that determines when the large model needs to correct the small model's inaccurate predictions. To evaluate our framework across different tasks and models, we apply BiLD to various text generation scenarios encompassing machine translation on IWSLT 2017 De-En and WMT 2014 De-En, and summarization on XSUM and CNN/DailyMail. On an NVIDIA T4 GPU, our framework achieves a speedup of up to 2.12x speedup with minimal generation quality degradation. Furthermore, our framework is fully plug-and-play and can be applied without any modifications in the training process or model architecture. Our code is open-sourced

We study the problem of learning goal-conditioned policies in Minecraft, a popular, widely accessible yet challenging open-ended environment for developing human-level multi-task agents. We first identify two main challenges of learning such policies: 1) the indistinguishability of tasks from the state distribution, due to the vast scene diversity, and 2) the non-stationary nature of environment dynamics caused by partial observability. To tackle the first challenge, we propose Goal-Sensitive Backbone (GSB) for the policy to encourage the emergence of goal-relevant visual state representations. To tackle the second challenge, the policy is further fueled by an adaptive horizon prediction module that helps alleviate the learning uncertainty brought by the non-stationary dynamics. Experiments on 20 Minecraft tasks show that our method significantly outperforms the best baseline so far; in many of them, we double the performance. Our ablation and exploratory studies then explain how our approach beat the counterparts and also unveil the surprising bonus of zero-shot generalization to new scenes (biomes). We hope our agent could help shed some light on learning goal-conditioned, multi-task agents in challenging, open-ended environments like Minecraft.

Abstract--Although it has been demonstrated that Natural Language Processing (NLP) algorithms are vulnerable to deliberate attacks, the question of whether such weaknesses can lead to software security threats is under-explored. To bridge this gap, we conducted vulnerability tests on Text-to-SQL systems that are commonly used to create natural language interfaces to databases. We showed that the Text-to-SQL modules within six commercial applications can be manipulated to produce malicious code, potentially leading to data breaches and Denial of Service attacks. This is the first demonstration that NLP (a) DoS attack: affecting the utility of one cloud server. In addition, experiments using four open-source language models verified that straightforward backdoor attacks on Text-to-SQL systems achieve a 100% success rate without affecting their performance. The aim of this work is to draw the community's attention to potential software security issues associated with NLP algorithms and encourage exploration of methods to mitigate against them.

Neural-symbolic computing (NeSy), which pursues the integration of the symbolic and statistical paradigms of cognition, has been an active research area of Artificial Intelligence (AI) for many years. As NeSy shows promise of reconciling the advantages of reasoning and interpretability of symbolic representation and robust learning in neural networks, it may serve as a catalyst for the next generation of AI. In the present paper, we provide a systematic overview of the recent developments and important contributions of NeSy research. Firstly, we introduce study history of this area, covering early work and foundations. We further discuss background concepts and identify key driving factors behind the development of NeSy. Afterward, we categorize recent landmark approaches along several main characteristics that underline this research paradigm, including neural-symbolic integration, knowledge representation, knowledge embedding, and functionality. Next, we briefly discuss the successful application of modern NeSy approaches in several domains. Then, we benchmark several NeSy methods on three representative application tasks. Finally, we identify the open problems together with potential future research directions. This survey is expected to help new researchers enter this rapidly evolving field and accelerate the progress towards data-and knowledge-driven AI.

We present a scalable approach for learning open-world object-goal navigation (ObjectNav) -- the task of asking a virtual robot (agent) to find any instance of an object in an unexplored environment (e.g., "find a sink"). Our approach is entirely zero-shot -- i.e., it does not require ObjectNav rewards or demonstrations of any kind. Instead, we train on the image-goal navigation (ImageNav) task, in which agents find the location where a picture (i.e., goal image) was captured. Specifically, we encode goal images into a multimodal, semantic embedding space to enable training semantic-goal navigation (SemanticNav) agents at scale in unannotated 3D environments (e.g., HM3D). After training, SemanticNav agents can be instructed to find objects described in free-form natural language (e.g., "sink", "bathroom sink", etc.) by projecting language goals into the same multimodal, semantic embedding space. As a result, our approach enables open-world ObjectNav. We extensively evaluate our agents on three ObjectNav datasets (Gibson, HM3D, and MP3D) and observe absolute improvements in success of 4.2% - 20.0% over existing zero-shot methods. For reference, these gains are similar or better than the 5% improvement in success between the Habitat 2020 and 2021 ObjectNav challenge winners. In an open-world setting, we discover that our agents can generalize to compound instructions with a room explicitly mentioned (e.g., "Find a kitchen sink") and when the target room can be inferred (e.g., "Find a sink and a stove").

Recent studies report that autoregressive language models can successfully solve many NLP tasks via zero- and few-shot learning paradigms, which opens up new possibilities for using the pre-trained language models. This paper introduces two autoregressive GPT-like models with 1.3 billion and 13 billion parameters trained on 60 languages from 25 language families using Wikipedia and Colossal Clean Crawled Corpus. We reproduce the GPT-3 architecture using GPT-2 sources and the sparse attention mechanism; Deepspeed and Megatron frameworks allow us to parallelize the training and inference steps effectively. The resulting models show performance on par with the recently released XGLM models by Facebook, covering more languages and enhancing NLP possibilities for low resource languages of CIS countries and Russian small nations. We detail the motivation for the choices of the architecture design, thoroughly describe the data preparation pipeline, and train five small versions of the model to choose the most optimal multilingual tokenization strategy. We measure the model perplexity in all covered languages and evaluate it on the wide spectre of multilingual tasks, including classification, generative, sequence labeling and knowledge probing. The models were evaluated with the zero-shot and few-shot methods. Furthermore, we compared the classification tasks with the state-of-the-art multilingual model XGLM. source code and the mGPT XL model are publicly released.

Large language models such as ChatGPT are increasingly explored in medical domains. However, the absence of standard guidelines for performance evaluation has led to methodological inconsistencies. This study aims to summarize the available evidence on evaluating ChatGPT's performance in medicine and provide direction for future research. We searched ten medical literature databases on June 15, 2023, using the keyword "ChatGPT". A total of 3520 articles were identified, of which 60 were reviewed and summarized in this paper and 17 were included in the meta-analysis. The analysis showed that ChatGPT displayed an overall integrated accuracy of 56% (95% CI: 51%-60%, I2 = 87%) in addressing medical queries. However, the studies varied in question resource, question-asking process, and evaluation metrics. Moreover, many studies failed to report methodological details, including the version of ChatGPT and whether each question was used independently or repeatedly. Our findings revealed that although ChatGPT demonstrated considerable potential for application in healthcare, the heterogeneity of the studies and insufficient reporting may affect the reliability of these results. Further well-designed studies with comprehensive and transparent reporting are needed to evaluate ChatGPT's performance in medicine.

Transformer-based large language models (Vaswani et al., 2017) pretrained with diverse tasks have demonstrated strong ability for in-context learning (ICL), that is, the pretrained models can answer new queries based on a few in-context demonstrations (see, e.g., Brown et al. (2020) and references thereafter). ICL is one of the key abilities contributing to the success of large language models, which allows pretrained models to solve multiple downstream tasks without updating their model parameters. However, the statistical foundation of ICL is still in its infancy. A recent line of research aims to quantify ICL by studying transformers pretrained on the linear regression task with a Gaussian prior (Garg et al., 2022; Akyürek et al., 2022; Li et al., 2023b; Raventós et al., 2023). Specifically, Garg et al. (2022); Akyürek et al. (2022); Li et al. (2023b) study the setting where transformers are pretrained in an online manner using independent linear regression tasks with the same Gaussian prior. They find that such a pretrained transformer can perform ICL on fresh linear regression tasks.