"Thinking is for Doing." Humans can infer other people's mental states from observations--an ability called Theory-of-Mind (ToM)--and subsequently act pragmatically on those inferences. Existing question answering benchmarks such as ToMi ask models questions to make inferences about beliefs of characters in a story, but do not test whether models can then use these inferences to guide their actions. We propose a new evaluation paradigm for large language models (LLMs): Thinking for Doing (T4D), which requires models to connect inferences about others' mental states to actions in social scenarios. Experiments on T4D demonstrate that LLMs such as GPT-4 and PaLM 2 seemingly excel at tracking characters' beliefs in stories, but they struggle to translate this capability into strategic action. Our analysis reveals the core challenge for LLMs lies in identifying the implicit inferences about mental states without being explicitly asked about as in ToMi, that lead to choosing the correct action in T4D. To bridge this gap, we introduce a zero-shot prompting framework, Foresee and Reflect (FaR), which provides a reasoning structure that encourages LLMs to anticipate future challenges and reason about potential actions. FaR boosts GPT-4's performance from 50% to 71% on T4D, outperforming other prompting methods such as Chain-of-Thought and Self-Ask. Moreover, FaR generalizes to diverse out-of-distribution story structures and scenarios that also require ToM inferences to choose an action, consistently outperforming other methods including few-shot in-context learning.

Multimodal information extraction (MIE) aims to extract structured information from unstructured multimedia content. Due to the diversity of tasks and settings, most current MIE models are task-specific and data-intensive, which limits their generalization to real-world scenarios with diverse task requirements and limited labeled data. To address these issues, we propose a novel multimodal question answering (MQA) framework to unify three MIE tasks by reformulating them into a unified span extraction and multi-choice QA pipeline. Extensive experiments on six datasets show that: 1) Our MQA framework consistently and significantly improves the performances of various off-the-shelf large multimodal models (LMM) on MIE tasks, compared to vanilla prompting. 2) In the zero-shot setting, MQA outperforms previous state-of-the-art baselines by a large margin. In addition, the effectiveness of our framework can successfully transfer to the few-shot setting, enhancing LMMs on a scale of 10B parameters to be competitive or outperform much larger language models such as ChatGPT and GPT-4. Our MQA framework can serve as a general principle of utilizing LMMs to better solve MIE and potentially other downstream multimodal tasks.

In order to understand the in-context learning phenomenon, recent works have adopted a stylized experimental framework and demonstrated that Transformers can learn gradient-based learning algorithms for various classes of real-valued functions. However, the limitations of Transformers in implementing learning algorithms, and their ability to learn other forms of algorithms are not well understood. Additionally, the degree to which these capabilities are confined to attention-based models is unclear. Furthermore, it remains to be seen whether the insights derived from these stylized settings can be extrapolated to pretrained Large Language Models (LLMs). In this work, we take a step towards answering these questions by demonstrating the following: (a) On a test-bed with a variety of Boolean function classes, we find that Transformers can nearly match the optimal learning algorithm for 'simpler' tasks, while their performance deteriorates on more 'complex' tasks. Additionally, we find that certain attention-free models perform (almost) identically to Transformers on a range of tasks. (b) When provided a teaching sequence, i.e. a set of examples that uniquely identifies a function in a class, we show that Transformers learn more sample-efficiently. Interestingly, our results show that Transformers can learn to implement two distinct algorithms to solve a single task, and can adaptively select the more sample-efficient algorithm depending on the sequence of in-context examples. (c) Lastly, we show that extant LLMs, e.g. LLaMA-2, GPT-4, can compete with nearest-neighbor baselines on prediction tasks that are guaranteed to not be in their training set.

Large language models (LLMs) have exploded in popularity due to their new generative capabilities that go far beyond prior state-of-the-art. These technologies are increasingly being leveraged in various domains such as law, finance, and medicine. However, these models carry significant computational challenges, especially the compute and energy costs required for inference. Inference energy costs already receive less attention than the energy costs of training LLMs -- despite how often these large models are called on to conduct inference in reality (e.g., ChatGPT). As these state-of-the-art LLMs see increasing usage and deployment in various domains, a better understanding of their resource utilization is crucial for cost-savings, scaling performance, efficient hardware usage, and optimal inference strategies. In this paper, we describe experiments conducted to study the computational and energy utilization of inference with LLMs. We benchmark and conduct a preliminary analysis of the inference performance and inference energy costs of different sizes of LLaMA -- a recent state-of-the-art LLM -- developed by Meta AI on two generations of popular GPUs (NVIDIA V100 \& A100) and two datasets (Alpaca and GSM8K) to reflect the diverse set of tasks/benchmarks for LLMs in research and practice. We present the results of multi-node, multi-GPU inference using model sharding across up to 32 GPUs. To our knowledge, our work is the one of the first to study LLM inference performance from the perspective of computational and energy resources at this scale.

Recent methods in text-to-3D leverage powerful pretrained diffusion models to optimize NeRF. Notably, these methods are able to produce high-quality 3D scenes without training on 3D data. Due to the open-ended nature of the task, most studies evaluate their results with subjective case studies and user experiments, thereby presenting a challenge in quantitatively addressing the question: How has current progress in Text-to-3D gone so far? In this paper, we introduce T$^3$Bench, the first comprehensive text-to-3D benchmark containing diverse text prompts of three increasing complexity levels that are specially designed for 3D generation. To assess both the subjective quality and the text alignment, we propose two automatic metrics based on multi-view images produced by the 3D contents. The quality metric combines multi-view text-image scores and regional convolution to detect quality and view inconsistency. The alignment metric uses multi-view captioning and Large Language Model (LLM) evaluation to measure text-3D consistency. Both metrics closely correlate with different dimensions of human judgments, providing a paradigm for efficiently evaluating text-to-3D models. The benchmarking results, shown in Fig. 1, reveal performance differences among six prevalent text-to-3D methods. Our analysis further highlights the common struggles for current methods on generating surroundings and multi-object scenes, as well as the bottleneck of leveraging 2D guidance for 3D generation. Our project page is available at: https://t3bench.com.

Instruction tuning has emerged as a crucial process for harnessing the capabilities of large language models (LLMs) by providing explicit task instructions, leading to improved performance in various tasks. However, prevalent text-to-text instruction tuning (TextTuning) methods suffer from limitations in generalization, robustness, and controllability due to the ambiguity and lack of explicit structure in tasks. In this paper, we propose JsonTuning, a novel structure-to-structure approach for instruction tuning. By leveraging the versatility and structured nature of JSON to represent tasks, JsonTuning enhances generalization by helping the model understand essential task elements and their relations, improves robustness by minimizing ambiguity, and increases controllability by providing explicit control over the output. We conduct a comprehensive comparative study with diverse language models and evaluation benchmarks. Experimental results show that JsonTuning outperforms TextTuning in various applications, showcasing improved performance, adaptability, robustness, and controllability. By overcoming the limitations of TextTuning, JsonTuning demonstrates significant potential for more effective and reliable LLMs capable of handling diverse scenarios.

Understanding how climate change affects us and learning about available solutions are key steps toward empowering individuals and communities to mitigate and adapt to it. As Large Language Models (LLMs) rise in popularity, it is necessary to assess their capability in this domain. In this study, we present a comprehensive evaluation framework, grounded in science communication principles, to analyze LLM responses to climate change topics. Our framework emphasizes both the presentational and epistemological adequacy of answers, offering a fine-grained analysis of LLM generations. Spanning 8 dimensions, our framework discerns up to 30 distinct issues in model outputs. The task is a real-world example of a growing number of challenging problems where AI can complement and lift human performance. We introduce a novel and practical protocol for scalable oversight that uses AI Assistance and relies on raters with relevant educational backgrounds. We evaluate several recent LLMs and conduct a comprehensive analysis of the results, shedding light on both the potential and the limitations of LLMs in the realm of climate communication.

Visual language reasoning requires a system to extract text or numbers from information-dense images like charts or plots and perform logical or arithmetic reasoning to arrive at an answer. To tackle this task, existing work relies on either (1) an end-to-end vision-language model trained on a large amount of data, or (2) a two-stage pipeline where a captioning model converts the image into text that is further read by another large language model to deduce the answer. However, the former approach forces the model to answer a complex question with one single step, and the latter approach is prone to inaccurate or distracting information in the converted text that can confuse the language model. In this work, we propose a dual-system for multi-step multimodal reasoning, which consists of a "System-1" step for visual information extraction and a "System-2" step for deliberate reasoning. Given an input, System-2 breaks down the question into atomic sub-steps, each guiding System-1 to extract the information required for reasoning from the image. Experiments on chart and plot datasets show that our method with a pre-trained System-2 module performs competitively compared to prior work on in- and out-of-distribution data. By fine-tuning the System-2 module (LLaMA-2 70B) on only a small amount of data on multi-step reasoning, the accuracy of our method is further improved and surpasses the best fully-supervised end-to-end approach by 5.7% and a pipeline approach with FlanPaLM (540B) by 7.5% on a challenging dataset with human-authored questions.

Spoken language understanding (SLU) typically includes two subtasks: intent detection and slot filling. Currently, it has achieved great success in high-resource languages, but it still remains challenging in low-resource languages due to the scarcity of labeled training data. Hence, there is a growing interest in zero-shot cross-lingual SLU. Despite of the success of existing zero-shot cross-lingual SLU models, most of them neglect to achieve the mutual guidance between intent and slots. To address this issue, we propose an Intra-Inter Knowledge Distillation framework for zero-shot cross-lingual Spoken Language Understanding (I$^2$KD-SLU) to model the mutual guidance. Specifically, we not only apply intra-knowledge distillation between intent predictions or slot predictions of the same utterance in different languages, but also apply inter-knowledge distillation between intent predictions and slot predictions of the same utterance. Our experimental results demonstrate that our proposed framework significantly improves the performance compared with the strong baselines and achieves the new state-of-the-art performance on the MultiATIS++ dataset, obtaining a significant improvement over the previous best model in overall accuracy.

Large language models (LLMs) are trained on massive internet corpora that often contain copyrighted content. This poses legal and ethical challenges for the developers and users of these models, as well as the original authors and publishers. In this paper, we propose a novel technique for unlearning a subset of the training data from a LLM, without having to retrain it from scratch. We evaluate our technique on the task of unlearning the Harry Potter books from the Llama2-7b model (a generative language model recently open-sourced by Meta). While the model took over 184K GPU-hours to pretrain, we show that in about 1 GPU hour of finetuning, we effectively erase the model's ability to generate or recall Harry Potter-related content, while its performance on common benchmarks (such as Winogrande, Hellaswag, arc, boolq and piqa) remains almost unaffected. We make our fine-tuned model publicly available on HuggingFace for community evaluation. To the best of our knowledge, this is the first paper to present an effective technique for unlearning in generative language models. Our technique consists of three main components: First, we use a reinforced model that is further trained on the target data to identify the tokens that are most related to the unlearning target, by comparing its logits with those of a baseline model. Second, we replace idiosyncratic expressions in the target data with generic counterparts, and leverage the model's own predictions to generate alternative labels for every token. These labels aim to approximate the next-token predictions of a model that has not been trained on the target data. Third, we finetune the model on these alternative labels, which effectively erases the original text from the model's memory whenever it is prompted with its context.