As Large Language Models (LLMs) play an increasingly pivotal role in natural language processing applications, their safety concerns become critical areas of NLP research. This paper presents Safety and Over-Defensiveness Evaluation (SODE) benchmark: a collection of diverse safe and unsafe prompts with carefully designed evaluation methods that facilitate systematic evaluation, comparison, and analysis over 'safety' and 'over-defensiveness.' With SODE, we study a variety of LLM defense strategies over multiple state-of-the-art LLMs, which reveals several interesting and important findings, such as (a) the widely popular 'self-checking' techniques indeed improve the safety against unsafe inputs, but this comes at the cost of extreme over-defensiveness on the safe inputs, (b) providing a safety instruction along with in-context exemplars (of both safe and unsafe inputs) consistently improves safety and also mitigates undue over-defensiveness of the models, (c) providing contextual knowledge easily breaks the safety guardrails and makes the models more vulnerable to generating unsafe responses. Overall, our work reveals numerous such critical findings that we believe will pave the way and facilitate further research in improving the safety of LLMs.
Large Language Models (LLMs) have achieved remarkable performance across a wide variety of natural language tasks; however, their large size makes their inference slow and computationally expensive. Focusing on this problem, we propose to instruction tune LLMs with additional explicit losses from the intermediate layers (LITE) and show that it enables these layers to acquire 'good' generation ability without affecting the generation ability of the final layer. We perform 'dynamic confidence-based early exiting' at token level from the intermediate layers which improves the efficiency of text generation without compromising the quality of the generation. We conduct comprehensive experiments by instruction tuning LLaMA-2 models on the Alpaca dataset and holistically evaluate on four different human-instruction test sets. We show that dynamic early exiting achieves consistent and considerable inference computation cost improvements (37.86% for 7B and 46.35% for 13B model) while maintaining the generation quality of the responses. We further conduct a thorough analysis of the results over several important aspects, such as comparing the semantic similarity of the outputs and dissecting the efficiency improvements by comparing the number of tokens generated in the output. In summary, our work contributes to improving the efficiency of LLM inference while maintaining the generation quality, a crucial step en route to enabling their widespread adoption.
Large Language Models (LLMs) have achieved remarkable performance across a wide variety of natural language tasks; however, their large size makes their inference slow and computationally expensive which poses a practical challenge for resource constrained real-world applications. Focusing on this problem, we propose to instruction tune LLMs in a way that enables intermediate layer decoding for efficiently generating text, but importantly without compromising the quality of the generation. Specifically, we instruction tune LLMs with additional explicit Losses from the InTermediate layErs (LITE) and show that it enables these layers to acquire 'good' generation ability without affecting the generation ability of the final layer. We perform 'dynamic confidence-based early exiting' at token level from the intermediate layers which improves the efficiency of inference while maintaining the generation quality. We conduct comprehensive experiments by instruction tuning LLaMA-2 models on the widely used Alpaca dataset and holistically evaluate on four different human-instruction test sets: Vicuna, WizardLM, Koala, and Self-Instruct. We show that 'dynamic early exiting' achieves consistent and considerable cost improvements (37.86% on average) while maintaining the generation quality of the responses. We further conduct a thorough analysis of the results over several important aspects, such as comparing the semantic similarity of the outputs and dissecting the efficiency improvements by comparing the number of tokens generated in the output. In summary, our work contributes to improving the efficiency of LLM inference while maintaining the generation quality, a crucial step en route to enabling their widespread adoption.
Logical reasoning is fundamental for humans yet presents a substantial challenge in the domain of Artificial Intelligence. Initially, researchers used Knowledge Representation and Reasoning (KR) systems that did not scale and required non trivial manual effort. Recently, the emergence of large language models (LLMs) has demonstrated the ability to overcome various limitations of formal Knowledge Representation (KR) systems. Consequently, there is a growing interest in using LLMs for logical reasoning via natural language. This work strives to understand the proficiency of LLMs in logical reasoning by offering a brief review of the latest progress in this area; with a focus on the logical reasoning datasets, tasks, and the methods adopted to utilize LLMs for reasoning. To offer a thorough analysis, we have compiled a benchmark titled LogiGLUE. This includes 24 varied datasets encompassing deductive, abductive, and inductive reasoning. We have standardized these datasets into Seq2Seq tasks to facilitate straightforward training and evaluation for future research. Utilizing LogiGLUE as a foundation, we have trained an instruction fine tuned language model, resulting in LogiT5. We study single task training, multi task training, and a chain of thought knowledge distillation fine tuning technique to assess the performance of model across the different logical reasoning categories. By this comprehensive process, we aim to shed light on the capabilities and potential pathways for enhancing logical reasoning proficiency in LLMs, paving the way for more advanced and nuanced developments in this critical field.
Though state-of-the-art (SOTA) NLP systems have achieved remarkable performance on a variety of language understanding tasks, they primarily focus on questions that have a correct and a definitive answer. However, in real-world applications, users often ask questions that don't have a definitive answer. Incorrectly answering such questions certainly hampers a system's reliability and trustworthiness. Can SOTA models accurately identify such questions and provide a reasonable response? To investigate the above question, we introduce QnotA, a dataset consisting of five different categories of questions that don't have definitive answers. Furthermore, for each QnotA instance, we also provide a corresponding QA instance i.e. an alternate question that ''can be'' answered. With this data, we formulate three evaluation tasks that test a system's ability to 'identify', 'distinguish', and 'justify' QnotA questions. Through comprehensive experiments, we show that even SOTA models including GPT-3 and Flan T5 do not fare well on these tasks and lack considerably behind the human performance baseline. We conduct a thorough analysis which further leads to several interesting findings. Overall, we believe our work and findings will encourage and facilitate further research in this important area and help develop more robust models.
Recently developed large language models have achieved remarkable success in generating fluent and coherent text. However, these models often tend to 'hallucinate' which critically hampers their reliability. In this work, we address this crucial problem and propose an approach that actively detects and mitigates hallucinations during the generation process. Specifically, we first identify the candidates of potential hallucination leveraging the model's logit output values, check their correctness through a validation procedure, mitigate the detected hallucinations, and then continue with the generation process. Through extensive experiments with the 'article generation task', we first demonstrate the individual efficacy of our detection and mitigation techniques. Specifically, the detection technique achieves a recall of 88% and the mitigation technique successfully mitigates 57.6% of the correctly detected hallucinations. Importantly, our mitigation technique does not introduce new hallucinations even in the case of incorrectly detected hallucinations, i.e., false positives. Then, we show that the proposed active detection and mitigation approach successfully reduces the hallucinations of the GPT-3 model from 47.5% to 14.5% on average. In summary, our work contributes to improving the reliability and trustworthiness of large language models, a crucial step en route to enabling their widespread adoption in real-world applications.
Pre-training on large corpora of text enables the language models to acquire a vast amount of factual and commonsense knowledge which allows them to achieve remarkable performance on a variety of language understanding tasks. They typically acquire this knowledge by learning from the pre-training text and capturing certain patterns from it. However, real-world settings often present scenarios that do not abide by these patterns i.e. scenarios that break the common assumptions. Can state-of-the-art NLP models correctly reason over the contexts of such scenarios? Addressing the above question, in this paper, we investigate the ability of models to correctly reason over contexts that break the common assumptions. To this end, we first systematically create evaluation data in which each data instance consists of (a) a common assumption, (b) a context that follows the assumption, (c) a context that breaks the assumption, and (d) questions based on the contexts. Then, through evaluations on multiple models including GPT-3 and Flan T5, we show that while doing fairly well on contexts that follow the common assumptions, the models struggle to correctly reason over contexts that break those assumptions. Specifically, the performance gap is as high as 20% absolute points. Furthermore, we thoroughly analyze these results revealing several interesting findings. We believe our work and findings will encourage and facilitate further research in developing more robust models that can also reliably reason over contexts that break the common assumptions. Data is available at \url{https://github.com/nrjvarshney/break_the_common_assumptions}.
State-of-the-art natural language processing models have been shown to achieve remarkable performance in 'closed-world' settings where all the labels in the evaluation set are known at training time. However, in real-world settings, 'novel' instances that do not belong to any known class are often observed. This renders the ability to deal with novelties crucial. To initiate a systematic research in this important area of 'dealing with novelties', we introduce 'NoveltyTask', a multi-stage task to evaluate a system's performance on pipelined novelty 'detection' and 'accommodation' tasks. We provide mathematical formulation of NoveltyTask and instantiate it with the authorship attribution task that pertains to identifying the correct author of a given text. We use Amazon reviews corpus and compile a large dataset (consisting of 250k instances across 200 authors/labels) for NoveltyTask. We conduct comprehensive experiments and explore several baseline methods for the task. Our results show that the methods achieve considerably low performance making the task challenging and leaving sufficient room for improvement. Finally, we believe our work will encourage research in this underexplored area of dealing with novelties, an important step en route to developing robust systems.
Despite remarkable progress made in natural language processing, even the state-of-the-art models often make incorrect predictions. Such predictions hamper the reliability of systems and limit their widespread adoption in real-world applications. 'Selective prediction' partly addresses the above concern by enabling models to abstain from answering when their predictions are likely to be incorrect. While selective prediction is advantageous, it leaves us with a pertinent question 'what to do after abstention'. To this end, we present an explorative study on 'Post-Abstention', a task that allows re-attempting the abstained instances with the aim of increasing 'coverage' of the system without significantly sacrificing its 'accuracy'. We first provide mathematical formulation of this task and then explore several methods to solve it. Comprehensive experiments on 11 QA datasets show that these methods lead to considerable risk improvements -- performance metric of the Post-Abstention task -- both in the in-domain and the out-of-domain settings. We also conduct a thorough analysis of these results which further leads to several interesting findings. Finally, we believe that our work will encourage and facilitate further research in this important area of addressing the reliability of NLP systems.
Learning to detect, characterize and accommodate novelties is a challenge that agents operating in open-world domains need to address to be able to guarantee satisfactory task performance. Certain novelties (e.g., changes in environment dynamics) can interfere with the performance or prevent agents from accomplishing task goals altogether. In this paper, we introduce general methods and architectural mechanisms for detecting and characterizing different types of novelties, and for building an appropriate adaptive model to accommodate them utilizing logical representations and reasoning methods. We demonstrate the effectiveness of the proposed methods in evaluations performed by a third party in the adversarial multi-agent board game Monopoly. The results show high novelty detection and accommodation rates across a variety of novelty types, including changes to the rules of the game, as well as changes to the agent's action capabilities.