Pegah Alipoormolabashi

Aarohi Srivastava, Abhinav Rastogi, Abhishek Rao et al. · allen-ai, amazon-science

Language models demonstrate both quantitative improvement and new qualitative capabilities with increasing scale. Despite their potentially transformative impact, these new capabilities are as yet poorly characterized. In order to inform future research, prepare for disruptive new model capabilities, and ameliorate socially harmful effects, it is vital that we understand the present and near-future capabilities and limitations of language models. To address this challenge, we introduce the Beyond the Imitation Game benchmark (BIG-bench). BIG-bench currently consists of 204 tasks, contributed by 450 authors across 132 institutions. Task topics are diverse, drawing problems from linguistics, childhood development, math, common-sense reasoning, biology, physics, social bias, software development, and beyond. BIG-bench focuses on tasks that are believed to be beyond the capabilities of current language models. We evaluate the behavior of OpenAI's GPT models, Google-internal dense transformer architectures, and Switch-style sparse transformers on BIG-bench, across model sizes spanning millions to hundreds of billions of parameters. In addition, a team of human expert raters performed all tasks in order to provide a strong baseline. Findings include: model performance and calibration both improve with scale, but are poor in absolute terms (and when compared with rater performance); performance is remarkably similar across model classes, though with benefits from sparsity; tasks that improve gradually and predictably commonly involve a large knowledge or memorization component, whereas tasks that exhibit "breakthrough" behavior at a critical scale often involve multiple steps or components, or brittle metrics; social bias typically increases with scale in settings with ambiguous context, but this can be improved with prompting.

Yizhong Wang, Swaroop Mishra, Pegah Alipoormolabashi et al. · allen-ai, amazon-science

How well can NLP models generalize to a variety of unseen tasks when provided with task instructions? To address this question, we first introduce Super-NaturalInstructions, a benchmark of 1,616 diverse NLP tasks and their expert-written instructions. Our collection covers 76 distinct task types, including but not limited to classification, extraction, infilling, sequence tagging, text rewriting, and text composition. This large and diverse collection of tasks enables rigorous benchmarking of cross-task generalization under instructions -- training models to follow instructions on a subset of tasks and evaluating them on the remaining unseen ones. Furthermore, we build Tk-Instruct, a transformer model trained to follow a variety of in-context instructions (plain language task definitions or k-shot examples). Our experiments show that Tk-Instruct outperforms existing instruction-following models such as InstructGPT by over 9% on our benchmark despite being an order of magnitude smaller. We further analyze generalization as a function of various scaling parameters, such as the number of observed tasks, the number of instances per task, and model sizes. We hope our dataset and model facilitate future progress towards more general-purpose NLP models.

Shikhar Singh, Nuan Wen, Yu Hou et al.

Commonsense reasoning is intuitive for humans but has been a long-term challenge for artificial intelligence (AI). Recent advancements in pretrained language models have shown promising results on several commonsense benchmark datasets. However, the reliability and comprehensiveness of these benchmarks towards assessing model's commonsense reasoning ability remains unclear. To this end, we introduce a new commonsense reasoning benchmark dataset comprising natural language true/false statements, with each sample paired with its complementary counterpart, resulting in 4k sentence pairs. We propose a pairwise accuracy metric to reliably measure an agent's ability to perform commonsense reasoning over a given situation. The dataset is crowdsourced and enhanced with an adversarial model-in-the-loop setup to incentivize challenging samples. To facilitate a systematic analysis of commonsense capabilities, we design our dataset along the dimensions of knowledge domains, reasoning scenarios and numeracy. Experimental results demonstrate that our strongest baseline (UnifiedQA-3B), after fine-tuning, achieves ~71% standard accuracy and ~51% pairwise accuracy, well below human performance (~95% for both metrics). The dataset is available at https://github.com/PlusLabNLP/Com2Sense.

Peter Zeng, Pegah Alipoormolabashi, Jihu Mun et al.

Responsible use of Authorship Verification (AV) systems not only requires high accuracy but also interpretable solutions. More importantly, for systems to be used to make decisions with real-world consequences requires the model's prediction to be explainable using interpretable features that can be traced to the original texts. Neural methods achieve high accuracies, but their representations lack direct interpretability. Furthermore, LLM predictions cannot be explained faithfully -- if there is an explanation given for a prediction, it doesn't represent the reasoning process behind the model's prediction. In this paper, we introduce Residualized Similarity (RS), a novel method that supplements systems using interpretable features with a neural network to improve their performance while maintaining interpretability. Authorship verification is fundamentally a similarity task, where the goal is to measure how alike two documents are. The key idea is to use the neural network to predict a similarity residual, i.e. the error in the similarity predicted by the interpretable system. Our evaluation across four datasets shows that not only can we match the performance of state-of-the-art authorship verification models, but we can show how and to what degree the final prediction is faithful and interpretable.

Biddut Sarker Bijoy, Mohammad Saqib Hasan, Pegah Alipoormolabashi et al.

Multi-agent systems with smaller language models (SLMs) present a viable alternative to single agent systems powered by large language models (LLMs) for addressing complex problems. In this work, we study how these alternatives compare in terms of both effectiveness and efficiency. To study this trade-off, we instantiate single and multi-agent systems for the complex problems in the AppWorld environment using different sized language models. We find that difficulties with long-trajectory learning in smaller language models (SLMs) limit their performance. Even when trained for specialized roles, SLMs fail to learn all subtasks effectively. To address this issue, we introduce a simple progressive sub-task training strategy, which introduces new sub-tasks progressively in each training epoch. We find that this novel strategy, analogous to instance level curriculum learning, consistently improves the effectiveness of multi-agents at all configurations. Our Pareto analysis shows that fine-tuned multi-agent systems yield better effectiveness-efficiency trade-offs. Additional ablations and analyses shows the importance of our progressive training strategy and its ability to reduce subtask error rates.

Pegah Alipoormolabashi, Ajay Patel, Niranjan Balasubramanian

Authorship misattribution can have profound consequences in real life. In forensic settings simply being considered as one of the potential authors of an evidential piece of text or communication can result in undesirable scrutiny. This raises a fairness question: Is every author in the candidate pool at equal risk of misattribution? Standard evaluation measures for authorship attribution systems do not explicitly account for this notion of fairness. We introduce a simple measure, Misattribution Unfairness Index (MAUIk), which is based on how often authors are ranked in the top k for texts they did not write. Using this measure we quantify the unfairness of five models on two different datasets. All models exhibit high levels of unfairness with increased risks for some authors. Furthermore, we find that this unfairness relates to how the models embed the authors as vectors in the latent search space. In particular, we observe that the risk of misattribution is higher for authors closer to the centroid (or center) of the embedded authors in the haystack. These results indicate the potential for harm and the need for communicating with and calibrating end users on misattribution risk when building and providing such models for downstream use.

Sayontan Ghosh, Mahnaz Koupaee, Yash Kumar Lal et al.

Understanding multiparty conversations demands robust Theory of Mind (ToM) capabilities, including the ability to track dynamic information, manage knowledge asymmetries, and distinguish relevant information across extended exchanges. To advance ToM evaluation in such settings, we present a carefully designed scalable methodology for generating high-quality benchmark conversation-question pairs with these characteristics. Using this methodology, we create $\texttt{DIAMONDs}$, a new conversational QA dataset covering common business, financial or other group interactions. In these goal-oriented conversations, participants often have to track certain numerical quantities (say $\textit{expected profit}$) of interest that can be derived from other variable quantities (like $\textit{marketing expenses, expected sales, salary}$, etc.), whose values also change over the course of the conversation. $\texttt{DIAMONDs}$ questions pose simple numerical reasoning problems over such quantities of interest (e.g., $\textit{funds required for charity events, expected company profit next quarter}$, etc.) in the context of the information exchanged in conversations. This allows for precisely evaluating ToM capabilities for carefully tracking and reasoning over participants' knowledge states. Our evaluation of state-of-the-art language models reveals significant challenges in handling participant-centric reasoning, specifically in situations where participants have false beliefs. Models also struggle with conversations containing distractors and show limited ability to identify scenarios with insufficient information. These findings highlight current models' ToM limitations in handling real-world multi-party conversations.

Te-Lin Wu, Alex Spangher, Pegah Alipoormolabashi et al.

The ability to sequence unordered events is an essential skill to comprehend and reason about real world task procedures, which often requires thorough understanding of temporal common sense and multimodal information, as these procedures are often communicated through a combination of texts and images. Such capability is essential for applications such as sequential task planning and multi-source instruction summarization. While humans are capable of reasoning about and sequencing unordered multimodal procedural instructions, whether current machine learning models have such essential capability is still an open question. In this work, we benchmark models' capability of reasoning over and sequencing unordered multimodal instructions by curating datasets from popular online instructional manuals and collecting comprehensive human annotations. We find models not only perform significantly worse than humans but also seem incapable of efficiently utilizing the multimodal information. To improve machines' performance on multimodal event sequencing, we propose sequentiality-aware pretraining techniques that exploit the sequential alignment properties of both texts and images, resulting in > 5% significant improvements.