We introduce HallusionBench, a comprehensive benchmark designed for the evaluation of image-context reasoning. This benchmark presents significant challenges to advanced large visual-language models (LVLMs), such as GPT-4V(Vision) and LLaVA-1.5, by emphasizing nuanced understanding and interpretation of visual data. The benchmark comprises 346 images paired with 1129 questions, all meticulously crafted by human experts. We introduce a novel structure for these visual questions designed to establish control groups. This structure enables us to conduct a quantitative analysis of the models' response tendencies, logical consistency, and various failure modes. In our evaluation on HallusionBench, we benchmarked 13 different models, highlighting a 31.42% question-pair accuracy achieved by the state-of-the-art GPT-4V. Notably, all other evaluated models achieve accuracy below 16%. Moreover, our analysis not only highlights the observed failure modes, including language hallucination and visual illusion, but also deepens an understanding of these pitfalls. Our comprehensive case studies within HallusionBench shed light on the challenges of hallucination and illusion in LVLMs. Based on these insights, we suggest potential pathways for their future improvement. The benchmark and codebase can be accessed at https://github.com/tianyi-lab/HallusionBench.

We study the problem of building a controller that can follow open-ended instructions in open-world environments. We propose to follow reference videos as instructions, which offer expressive goal specifications while eliminating the need for expensive text-gameplay annotations. A new learning framework is derived to allow learning such instruction-following controllers from gameplay videos while producing a video instruction encoder that induces a structured goal space. We implement our agent GROOT in a simple yet effective encoder-decoder architecture based on causal transformers. We evaluate GROOT against open-world counterparts and human players on a proposed Minecraft SkillForge benchmark. The Elo ratings clearly show that GROOT is closing the human-machine gap as well as exhibiting a 70% winning rate over the best generalist agent baseline. Qualitative analysis of the induced goal space further demonstrates some interesting emergent properties, including the goal composition and complex gameplay behavior synthesis. The project page is available at https://craftjarvis-groot.github.io.

As large language models (LLMs) become more capable, there is an urgent need for interpretable and transparent tools. Current methods are difficult to implement, and accessible tools to analyze model internals are lacking. To bridge this gap, we present DeepDecipher - an API and interface for probing neurons in transformer models' MLP layers. DeepDecipher makes the outputs of advanced interpretability techniques for LLMs readily available. The easy-to-use interface also makes inspecting these complex models more intuitive. This paper outlines DeepDecipher's design and capabilities. We demonstrate how to analyze neurons, compare models, and gain insights into model behavior. For example, we contrast DeepDecipher's functionality with similar tools like Neuroscope and OpenAI's Neuron Explainer. DeepDecipher enables efficient, scalable analysis of LLMs. By granting access to state-of-the-art interpretability methods, DeepDecipher makes LLMs more transparent, trustworthy, and safe. Researchers, engineers, and developers can quickly diagnose issues, audit systems, and advance the field.

We investigate the role of various demonstration components in the in-context learning (ICL) performance of large language models (LLMs). Specifically, we explore the impacts of ground-truth labels, input distribution, and complementary explanations, particularly when these are altered or perturbed. We build on previous work, which offers mixed findings on how these elements influence ICL. To probe these questions, we employ explainable NLP (XNLP) methods and utilize saliency maps of contrastive demonstrations for both qualitative and quantitative analysis. Our findings reveal that flipping ground-truth labels significantly affects the saliency, though it's more noticeable in larger LLMs. Our analysis of the input distribution at a granular level reveals that changing sentiment-indicative terms in a sentiment analysis task to neutral ones does not have as substantial an impact as altering ground-truth labels. Finally, we find that the effectiveness of complementary explanations in boosting ICL performance is task-dependent, with limited benefits seen in sentiment analysis tasks compared to symbolic reasoning tasks. These insights are critical for understanding the functionality of LLMs and guiding the development of effective demonstrations, which is increasingly relevant in light of the growing use of LLMs in applications such as ChatGPT. Our research code is publicly available at https://github.com/paihengxu/XICL.

Contrastive pretraining of image-text foundation models, such as CLIP, demonstrated excellent zero-shot performance and improved robustness on a wide range of downstream tasks. However, these models utilize large transformer-based encoders with significant memory and latency overhead which pose challenges for deployment on mobile devices. In this work, we introduce MobileCLIP -- a new family of efficient image-text models optimized for runtime performance along with a novel and efficient training approach, namely multi-modal reinforced training. The proposed training approach leverages knowledge transfer from an image captioning model and an ensemble of strong CLIP encoders to improve the accuracy of efficient models. Our approach avoids train-time compute overhead by storing the additional knowledge in a reinforced dataset. MobileCLIP sets a new state-of-the-art latency-accuracy tradeoff for zero-shot classification and retrieval tasks on several datasets. Our MobileCLIP-S2 variant is 2.3$\times$ faster while more accurate compared to previous best CLIP model based on ViT-B/16. We further demonstrate the effectiveness of our multi-modal reinforced training by training a CLIP model based on ViT-B/16 image backbone and achieving +2.9% average performance improvement on 38 evaluation benchmarks compared to the previous best. Moreover, we show that the proposed approach achieves 10$\times$-1000$\times$ improved learning efficiency when compared with non-reinforced CLIP training.

In this work, we present a novel method to tackle the token generation challenge in Vision Language Models (VLMs) for video and image understanding, called LLaMA-VID. Current VLMs, while proficient in tasks like image captioning and visual question answering, face computational burdens when processing long videos due to the excessive visual tokens. LLaMA-VID addresses this issue by representing each frame with two distinct tokens, namely context token and content token. The context token encodes the overall image context based on user input, whereas the content token encapsulates visual cues in each frame. This dual-token strategy significantly reduces the overload of long videos while preserving critical information. Generally, LLaMA-VID empowers existing frameworks to support hour-long videos and pushes their upper limit with an extra context token. It is proved to surpass previous methods on most of video- or image-based benchmarks. Code is available https://github.com/dvlab-research/LLaMA-VID}{https://github.com/dvlab-research/LLaMA-VID

Artificial intelligence and machine learning have significantly bolstered the technological world. This paper explores the potential of transfer learning in natural language processing focusing mainly on sentiment analysis. The models trained on the big data can also be used where data are scarce. The claim is that, compared to training models from scratch, transfer learning, using pre-trained BERT models, can increase sentiment classification accuracy. The study adopts a sophisticated experimental design that uses the IMDb dataset of sentimentally labelled movie reviews. Pre-processing includes tokenization and encoding of text data, making it suitable for NLP models. The dataset is used on a BERT based model, measuring its performance using accuracy. The result comes out to be 100 per cent accurate. Although the complete accuracy could appear impressive, it might be the result of overfitting or a lack of generalization. Further analysis is required to ensure the model's ability to handle diverse and unseen data. The findings underscore the effectiveness of transfer learning in NLP, showcasing its potential to excel in sentiment analysis tasks. However, the research calls for a cautious interpretation of perfect accuracy and emphasizes the need for additional measures to validate the model's generalization.

Recent years have seen a surge in the popularity of commercial AI products based on generative, multi-purpose AI systems promising a unified approach to building machine learning (ML) models into technology. However, this ambition of "generality" comes at a steep cost to the environment, given the amount of energy these systems require and the amount of carbon that they emit. In this work, we propose the first systematic comparison of the ongoing inference cost of various categories of ML systems, covering both task-specific (i.e. finetuned models that carry out a single task) and `general-purpose' models, (i.e. those trained for multiple tasks). We measure deployment cost as the amount of energy and carbon required to perform 1,000 inferences on representative benchmark dataset using these models. We find that multi-purpose, generative architectures are orders of magnitude more expensive than task-specific systems for a variety of tasks, even when controlling for the number of model parameters. We conclude with a discussion around the current trend of deploying multi-purpose generative ML systems, and caution that their utility should be more intentionally weighed against increased costs in terms of energy and emissions. All the data from our study can be accessed via an interactive demo to carry out further exploration and analysis.

Large Language Models (LLMs) are notorious for blending fact with fiction and generating non-factual content, known as hallucinations. To tackle this challenge, we propose an interactive system that helps users obtain insights into the reliability of the generated text. Our approach is based on the idea that the self-consistency of multiple samples generated by the same LLM relates to its confidence in individual claims in the generated texts. Using this idea, we design RELIC, an interactive system that enables users to investigate and verify semantic-level variations in multiple long-form responses. This allows users to recognize potentially inaccurate information in the generated text and make necessary corrections. From a user study with ten participants, we demonstrate that our approach helps users better verify the reliability of the generated text. We further summarize the design implications and lessons learned from this research for inspiring future studies on reliable human-LLM interactions.

The overwhelming success of OpenAI's ChatGPT, whose first version was released one year ago, has led to an undeniable surge of excitement about large language models (LLMs) among researchers and the general public alike. OpenAI's anything-but-open approach to sharing its models or information about training them, however, has led to an equally passionate reaction among those who feel that AI development should be widely accessible and that data usage should be transparent in order to protect the rights of those who have contributed the data and that data - a resource crucial to the development and understanding of AI models - should be shared with the broader research community. The call for transparency has begun to bear fruit. High-profile language models like Falcon,[Almazrouei et al., 2023] LLaMa2 [Touvron et al., 2023] and MPT [MosaicML NLP Team, 2023] - to name just a few - come very close to a classic definition of open source. A central part of OpenLLM France's mission is to contribute to this momentum by building language models and remaining fully transparent about every step of model training, including the data used for training. Another objective, which we find equally important, is to increase the availability of language models and training data geared to languages other than English and to non-anglophone cultures. Indeed, the majority of the high-profile LLMs available today are trained primarily on English documents coming from anglophone cultures. Only 0.16% of the data used to train LLaMa2 comes from French, for example.