Image generation models are poised to become ubiquitous in a range of applications. These models are often fine-tuned and evaluated using human quality judgments that assume a universal standard, failing to consider the subjectivity of such tasks. To investigate how to quantify subjectivity, and the scale of its impact, we measure how assessments differ among human annotators across different use cases. Simulating the effects of ordinarily latent elements of annotators subjectivity, we contrive a set of motivations (t-shirt graphics, presentation visuals, and phone background images) to contextualize a set of crowdsourcing tasks. Our results show that human evaluations of images vary within individual contexts and across combinations of contexts. Three key factors affecting this subjectivity are image appearance, image alignment with text, and representation of objects mentioned in the text. Our study highlights the importance of taking individual users and contexts into account, both when building and evaluating generative models

Robotic grasping is a fundamental aspect of robot functionality, defining how robots interact with objects. Despite substantial progress, its generalizability to counter-intuitive or long-tailed scenarios, such as objects with uncommon materials or shapes, remains a challenge. In contrast, humans can easily apply their intuitive physics to grasp skillfully and change grasps efficiently, even for objects they have never seen before. This work delves into infusing such physical commonsense reasoning into robotic manipulation. We introduce PhyGrasp, a multimodal large model that leverages inputs from two modalities: natural language and 3D point clouds, seamlessly integrated through a bridge module. The language modality exhibits robust reasoning capabilities concerning the impacts of diverse physical properties on grasping, while the 3D modality comprehends object shapes and parts. With these two capabilities, PhyGrasp is able to accurately assess the physical properties of object parts and determine optimal grasping poses. Additionally, the model's language comprehension enables human instruction interpretation, generating grasping poses that align with human preferences. To train PhyGrasp, we construct a dataset PhyPartNet with 195K object instances with varying physical properties and human preferences, alongside their corresponding language descriptions. Extensive experiments conducted in the simulation and on the real robots demonstrate that PhyGrasp achieves state-of-the-art performance, particularly in long-tailed cases, e.g., about 10% improvement in success rate over GraspNet. Project page: https://sites.google.com/view/phygrasp

This work explores whether language models encode meaningfully grounded representations of sounds of objects. We learn a linear probe that retrieves the correct text representation of an object given a snippet of audio related to that object, where the sound representation is given by a pretrained audio model. This probe is trained via a contrastive loss that pushes the language representations and sound representations of an object to be close to one another. After training, the probe is tested on its ability to generalize to objects that were not seen during training. Across different language models and audio models, we find that the probe generalization is above chance in many cases, indicating that despite being trained only on raw text, language models encode grounded knowledge of sounds for some objects.

Knowledge-intensive language tasks (KILTs) typically require retrieving relevant documents from trustworthy corpora, e.g., Wikipedia, to produce specific answers. Very recently, a pre-trained generative retrieval model for KILTs, named CorpusBrain, was proposed and reached new state-of-the-art retrieval performance. However, most existing research on KILTs, including CorpusBrain, has predominantly focused on a static document collection, overlooking the dynamic nature of real-world scenarios, where new documents are continuously being incorporated into the source corpus. To address this gap, it is crucial to explore the capability of retrieval models to effectively handle the dynamic retrieval scenario inherent in KILTs. In this work, we first introduce the continual document learning (CDL) task for KILTs and build a novel benchmark dataset named KILT++ based on the original KILT dataset for evaluation. Then, we conduct a comprehensive study over the use of pre-trained CorpusBrain on KILT++. Unlike the promising results in the stationary scenario, CorpusBrain is prone to catastrophic forgetting in the dynamic scenario, hence hampering the retrieval performance. To alleviate this issue, we propose CorpusBrain++, a continual generative pre-training framework. Empirical results demonstrate the significant effectiveness and remarkable efficiency of CorpusBrain++ in comparison to both traditional and generative IR methods.

Clinical trials require numerous documents to be written -- protocols, consent forms, clinical study reports and others. Large language models (LLMs) offer the potential to rapidly generate first versions of these documents, however there are concerns about the quality of their output. Here we report an evaluation of LLMs in generating parts of one such document, clinical trial protocols. We find that an offthe-shelf LLM delivers reasonable results, especially when assessing content relevance and the correct use of terminology. However, deficiencies remain: specifically clinical thinking and logic, and appropriate use of references. To improve performance, we used retrieval-augmented generation (RAG) to prompt an LLM with accurate up-to-date information. As a result of using RAG, the writing quality of the LLM improves substantially, which has implications for the practical useability of LLMs in clinical trial-related writing.

Recent advancements in large language models (LLMs) have revealed their potential for achieving autonomous agents possessing human-level intelligence. However, existing benchmarks for evaluating LLM Agents either use static datasets, potentially leading to data leakage or focus only on single-agent scenarios, overlooking the complexities of multi-agent interactions. There is a lack of a benchmark that evaluates the diverse capabilities of LLM agents in multi-agent, dynamic environments. To this end, we introduce LLMArena, a novel and easily extensible framework for evaluating the diverse capabilities of LLM in multi-agent dynamic environments. LLMArena encompasses seven distinct gaming environments, employing Trueskill scoring to assess crucial abilities in LLM agents, including spatial reasoning, strategic planning, numerical reasoning, risk assessment, communication, opponent modeling, and team collaboration. We conduct an extensive experiment and human evaluation among different sizes and types of LLMs, showing that LLMs still have a significant journey ahead in their development towards becoming fully autonomous agents, especially in opponent modeling and team collaboration. We hope LLMArena could guide future research towards enhancing these capabilities in LLMs, ultimately leading to more sophisticated and practical applications in dynamic, multi-agent settings. The code and data will be available.

Methods for watermarking large language models have been proposed that distinguish AI-generated text from human-generated text by slightly altering the model output distribution, but they also distort the quality of the text, exposing the watermark to adversarial detection. More recently, distortion-free watermarking methods were proposed that require a secret key to detect the watermark. The prior methods generally embed zero-bit watermarks that do not provide additional information beyond tagging a text as being AI-generated. We extend an existing zero-bit distortion-free watermarking method by embedding multiple bits of meta-information as part of the watermark. We also develop a computationally efficient decoder that extracts the embedded information from the watermark with low bit error rate.

This study evaluates three state-of-the-art MLLMs -- GPT-4V, Gemini Pro, and the open-source model IDEFICS -- on the compositional natural language vision reasoning task NLVR. Given a human-written sentence paired with a synthetic image, this task requires the model to determine the truth value of the sentence with respect to the image. Despite the strong performance demonstrated by these models, we observe they perform poorly on NLVR, which was constructed to require compositional and spatial reasoning, and to be robust for semantic and systematic biases.

In this paper we explore the challenges of measuring sentiment in relation to Environmental, Social and Governance (ESG) social media. ESG has grown in importance in recent years with a surge in interest from the financial sector and the performance of many businesses has become based in part on their ESG related reputations. The use of sentiment analysis to measure ESG related reputation has developed and with it interest in the use of machines to do so. The era of digital media has created an explosion of new media sources, driven by the growth of social media platforms. This growing data environment has become an excellent source for behavioural insight studies across many disciplines that includes politics, healthcare and market research. Our study seeks to compare human performance with the cutting edge in machine performance in the measurement of ESG related sentiment. To this end researchers classify the sentiment of 150 tweets and a reliability measure is made. A gold standard data set is then established based on the consensus of 3 researchers and this data set is then used to measure the performance of different machine approaches: one based on the VADER dictionary approach to sentiment classification and then multiple language model approaches, including Llama2, T5, Mistral, Mixtral, FINBERT, GPT3.5 and GPT4.

Binary code representation learning has shown significant performance in binary analysis tasks. But existing solutions often have poor transferability, particularly in few-shot and zero-shot scenarios where few or no training samples are available for the tasks. To address this problem, we present CLAP (Contrastive Language-Assembly Pre-training), which employs natural language supervision to learn better representations of binary code (i.e., assembly code) and get better transferability. At the core, our approach boosts superior transfer learning capabilities by effectively aligning binary code with their semantics explanations (in natural language), resulting a model able to generate better embeddings for binary code. To enable this alignment training, we then propose an efficient dataset engine that could automatically generate a large and diverse dataset comprising of binary code and corresponding natural language explanations. We have generated 195 million pairs of binary code and explanations and trained a prototype of CLAP. The evaluations of CLAP across various downstream tasks in binary analysis all demonstrate exceptional performance. Notably, without any task-specific training, CLAP is often competitive with a fully supervised baseline, showing excellent transferability. We release our pre-trained model and code at https://github.com/Hustcw/CLAP.