Natural language is an instrument of interaction, where interlocutors produce and comprehend language to relay information to accomplish their intents. Both opportunities and challenges arise from the interactive nature of language. The response of participants to the language they comprehend can form a strong learning signal: did I achieve my intent? In the first part, I will define a simple, and general framework to rapidly learn from interaction feedback, and show how to use this signal to learn to produce natural language instructions. I will then discuss the problem of language-conditioned reinforcement learning, where benchmark development has been hindered because computing rewards requires resolving language semantics. I will describe a new approach to address this challenge.

Natural Language Learning via Interaction

In times of crisis, the human mind is often a voracious information forager. It might not be immediately apparent what one wants or needs, and people frequently look for answers to their most pressing questions and worst fears. In that context, the pandemic has demonstrated that social media sources, like erstwhile Twitter, are a rich medium for data-driven communication between experts and the public. However, as lay users, we must find needles in a haystack to distinguish credible and actionable information signals from the noise. In this work, we leverage the literature on crisis communication to propose an AI-driven sensemaking model that bridges the gap between what people seek and what they need during a crisis. Our model learns to contrast social media messages concerning expert guidance with subjective opinion and enables semantic interpretation of message characteristics based on the communicative intent of the message author. We provide examples from our tweet collection and present a hypothetical social media usage scenario to demonstrate the efficacy of our proposed model.

Sensemaking of Socially-Mediated Crisis Information

In the realm of research support tools, there exists a notable void in resources tailored specifically for aiding researchers during the crucial ideation phase of the research life-cycle. We address this gap by introducing 'Acceleron', a `Co-Pilot' for researchers, designed specifically to accelerate the ideation phase of the research life-cycle. Leveraging the reasoning and domain-specific skills of Large Language Models (LLMs) within an agent-based architecture with distinct personas, Acceleron aids researchers through the formulation of a comprehensive research proposals. It emulates the ideation process, engaging researchers in an interactive fashion to validate the novelty of the proposal and generate plausible set-of hypotheses. Notably, it addresses challenges inherent in LLMs, such as hallucinations, implements a two-stage aspect-based retrieval to manage precision-recall trade-offs, and tackles issues of unanswerability. Our observations and end-user evaluations illustrate the efficacy of Acceleron as an enhancer of researcher's productivity.

An Interactive Co-Pilot for Accelerated Research Ideation

With the rapid proliferation of artificial intelligence, there is growing concern over its potential to exacerbate existing biases and societal disparities and introduce novel ones. This issue has prompted widespread attention from academia, policymakers, industry, and civil society. While evidence suggests that integrating human perspectives can mitigate bias-related issues in AI systems, it also introduces challenges associated with cognitive biases inherent in human decision-making. Our research focuses on reviewing existing methodologies and ongoing investigations aimed at understanding annotation attributes that contribute to bias.

Blind Spots and Biases: Exploring the Role of Annotator Cognitive Biases in NLP

Interpretability tools that offer explanations in the form of a dialogue have demonstrated their efficacy in enhancing users' understanding (Slack et al., 2023; Shen et al., 2023), as one-off explanations may fall short in providing sufficient information to the user. Current solutions for dialogue-based explanations, however, often require external tools and modules and are not easily transferable to tasks they were not designed for. With LLMCheckup, we present an easily accessible tool that allows users to chat with any state-of-the-art large language model (LLM) about its behavior. We enable LLMs to generate explanations and perform user intent recognition without fine-tuning, by connecting them with a broad spectrum of Explainable AI (XAI) methods, including white-box explainability tools such as feature attributions, and self-explanations (e.g., for rationale generation). LLM-based (self-)explanations are presented as an interactive dialogue that supports follow-up questions and generates suggestions. LLMCheckup provides tutorials for operations available in the system, catering to individuals with varying levels of expertise in XAI and supporting multiple input modalities. We introduce a new parsing strategy that substantially enhances the user intent recognition accuracy of the LLM. Finally, we showcase for the tasks of fact checking and commonsense question answering.

LLMCheckup: Conversational Examination of Large Language Models via Interpretability Tools and Self-Explanations

When users want to write a story with a language model (LM) assistant, it is often very difficult to provide a prompt that clearly specifies all their interests. For the developers of LM assistants, it is also difficult to ensure their output stories come from a dataset without copyright concerns. Motivated by these limitations, we propose a coarse-to-fine (C2F) tree-based story generation framework, which is called C2F-StoryTree, where the LM iteratively generates more and more specific story prompts based on a user's input prompt and the desired plot selected by the user. To realize our C2F-StoryTree framework, we propose an entailment hierarchy (EH) text structure, in which a more specific response entails more general prompt (e.g., a story entails a summary). We also propose novel annotation tasks, decoding methods, and a human-and-machine-in-the-loop procedure to minimize the annotation cost of building the text structure. We build an entailment hierarchy dataset on top of the story datasets with desired licenses and styles, on which the developers can fine-tune or evaluate their LMs. In our experiments, we demonstrate that our C2F-StoryTree system not only allows users to collaborate with a LM in a new paradigm, but also generates the short stories with human-level quality. Furthermore, our entailment hierarchy and C2F generation substantially make the generated stories more diverse, coherent, engaging, and creative.

Fine-to-Coarse Entailment Hierarchy Construction for Coarse-to-Fine Story Generation

Benchmarks of the multilingual capabilities of text-to-image (T2I) models compare generated images prompted in a test language to an expected image distribution over a concept set. One such benchmark, ``Conceptual Coverage Across Languages'' (CoCo-CroLa), assesses the tangible noun inventory of T2I models by prompting them to generate pictures from a concept list translated to seven languages and comparing the output image populations. Unfortunately, we find that this benchmark contains translation errors of varying severity in Spanish, Japanese, and Chinese. We provide corrections for these errors and analyze how impactful they are on the utility and validity of CoCo-CroLa as a benchmark. We reassess multiple baseline T2I models with the revisions, compare the outputs elicited under the new translations to those conditioned on the old, and show that a correction's impactfulness on the image-domain benchmark results can be predicted in the text domain with similarity scores. Our findings will guide the future development of T2I multilinguality metrics by providing analytical tools for practical translation decisions.

Lost in Translation? Translation Errors and Challenges for Fair Assessment of Text-to-Image Models on Multilingual Concepts

We present FastFit, a Python package designed to provide fast and accurate few-shot classification, especially for scenarios with many semantically similar classes. FastFit utilizes a novel approach integrating batch contrastive learning and token-level similarity score. Compared to existing few-shot learning packages, such as SetFit, Transformers, or few-shot prompting of large language models via API calls, FastFit significantly improves multi-class classification performance in speed and accuracy across various English and Multilingual datasets. FastFit demonstrates a 3-20x improvement in training speed, completing training in just a few seconds. The FastFit package is now available on GitHub, presenting a user-friendly solution for NLP practitioners.

FastFit: Fast and Effective Few-Shot Text Classification with a Multitude of Classes

Our research integrates graph data with Large Language Models (LLMs), which, despite their advancements in various fields using large text corpora, face limitations in encoding entire graphs due to context size constraints. This paper introduces a new approach to encoding a graph with diverse modalities, such as text, image, and motif, coupled with prompts to approximate a graph's global connectivity, thereby enhancing LLMs' efficiency in processing complex graph structures. The study also presents GraphTMI, a novel benchmark for evaluating LLMs in graph structure analysis, focusing on homophily, motif presence, and graph difficulty. Key findings indicate that the image modality, especially with vision-language models like GPT-4V, is superior to text in balancing token limits and preserving essential information and comes close to prior graph neural net (GNN) encoders. Furthermore, the research assesses how various factors affect the performance of each encoding modality and outlines the existing challenges and potential future developments for LLMs in graph understanding and reasoning tasks. Our code and data are publicly available on our project page - https://minnesotanlp.github.io/GraphLLM/

Which Modality should I use - Text, Motif, or Image? : Understanding Graphs with Large Language Models

Are multimodal inputs necessary for grammar induction? Recent work has shown that multimodal training inputs can improve grammar induction. However, these improvements are based on comparisons to weak text-only baselines that were trained on relatively little textual data. To determine whether multimodal inputs are needed in regimes with large amounts of textual training data, we design a stronger text-only baseline, which we refer to as LC-PCFG. LC-PCFG is a C-PFCG that incorporates embeddings from text-only large language models (LLMs). We use a fixed grammar family to directly compare LC-PCFG to various multimodal grammar induction methods. We compare performance on four benchmark datasets. LC-PCFG provides an up to 17% relative improvement in Corpus-F1 compared to state-of-the-art multimodal grammar induction methods. LC-PCFG is also more computationally efficient, providing an up to 85% reduction in parameter count and $8.8\times$ reduction in training time compared to multimodal approaches. These results suggest that multimodal inputs may not be necessary for grammar induction, and emphasize the importance of strong vision-free baselines for evaluating the benefit of multimodal approaches.

Re-evaluating the Need for Visual Signals in Unsupervised Grammar Induction

Referring Expression Generation (REG) is the task of generating a description that unambiguously identifies a given target in the scene. Different from Image Captioning (IC), REG requires learning fine-grained characteristics of not only the scene objects but also their surrounding context. Referring expressions are usually not singular; an object can often be uniquely referenced in numerous ways, for instance, by color, by location, or by relationship with other objects. Most prior works, however, have not explored this 'aspect-based multiplicity' of referring expressions. Hence, in this work, we focus on the Aspect-Controlled REG task, which requires generating a referring expression conditioned on the input aspect(s), where an aspect captures a style of reference. By changing the input aspect such as color, location, action etc., one can generate multiple distinct expressions per target region. To solve this new task, we first modify BLIP for aligning image-regions and text-expressions. We achieve this through a novel approach for feeding the input by drawing a bounding box around the target image-region and prompting the model to generate the referring expression. Our base REG model already beats all prior works in CIDEr score. To tackle Aspect-Controlled REG, we append `aspect tokens' to the prompt and show that distinct expressions can be generated by just changing the prompt. Finally, to prove the high-quality and diversity of the data generated by our proposed aspect-controlled REG model, we also perform data-augmentation-based evaluation on the downstream Referring Expression Comprehension (REC) task. With just half of the real data augmented with the generated synthetic data, we achieve performance comparable to training with 100\% of real data, using a SOTA REC model.

Prompting Vision-Language Models For Aspect-Controlled Generation of Referring Expressions

CLIP-based classifiers rely on the prompt containing a {class name} that is known to the text encoder. Therefore, they perform poorly on new classes or the classes whose names rarely appear on the Internet (e.g., scientific names of birds). For fine-grained classification, we propose PEEB – an explainable and editable classifier to (1) express the class name into a set of text descriptors that describe the visual parts of that class; and (2) match the embeddings of the detected parts to their textual descriptors in each class to compute a logit score for classification. In a zero-shot setting where the class names are unknown, PEEB outperforms CLIP by a huge margin (∼10× in top-1 accuracy). Compared to part-based classifiers, PEEB is not only the state-of-the-art (SOTA) on the supervised-learning setting (88.80% and 92.20% accuracy on CUB-200 and Stanford Dogs-120, respectively) but also the first to enable users to edit the text descriptors to form a new classifier without any re-training. Compared to concept bottleneck models, PEEB is also the SOTA in both zero-shot and supervised-learning settings.

Underline digital video library

Natural Language Learning via Interaction

Sensemaking of Socially-Mediated Crisis Information

An Interactive Co-Pilot for Accelerated Research Ideation

Blind Spots and Biases: Exploring the Role of Annotator Cognitive Biases in NLP

LLMCheckup: Conversational Examination of Large Language Models via Interpretability Tools and Self-Explanations

Fine-to-Coarse Entailment Hierarchy Construction for Coarse-to-Fine Story Generation

Lost in Translation? Translation Errors and Challenges for Fair Assessment of Text-to-Image Models on Multilingual Concepts

FastFit: Fast and Effective Few-Shot Text Classification with a Multitude of Classes

Which Modality should I use - Text, Motif, or Image? : Understanding Graphs with Large Language Models

Re-evaluating the Need for Visual Signals in Unsupervised Grammar Induction

Prompting Vision-Language Models For Aspect-Controlled Generation of Referring Expressions

PEEB: Part-based Image Classifiers with an Explainable and Editable Language Bottleneck

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES