Amos Azaria


2023

pdf bib
The Internal State of an LLM Knows When It’s Lying
Amos Azaria | Tom Mitchell
Findings of the Association for Computational Linguistics: EMNLP 2023

While Large Language Models (LLMs) have shown exceptional performance in various tasks, one of their most prominent drawbacks is generating inaccurate or false information with a confident tone. In this paper, we provide evidence that the LLM’s internal state can be used to reveal the truthfulness of statements. This includes both statements provided to the LLM, and statements that the LLM itself generates. Our approach is to train a classifier that outputs the probability that a statement is truthful, based on the hidden layer activations of the LLM as it reads or generates the statement. Experiments demonstrate that given a set of test sentences, of which half are true and half false, our trained classifier achieves an average of 71% to 83% accuracy labeling which sentences are true versus false, depending on the LLM base model. Furthermore, we explore the relationship between our classifier’s performance and approaches based on the probability assigned to the sentence by the LLM. We show that while LLM-assigned sentence probability is related to sentence truthfulness, this probability is also dependent on sentence length and the frequencies of words in the sentence, resulting in our trained classifier providing a more reliable approach to detecting truthfulness, highlighting its potential to enhance the reliability of LLM-generated content and its practical applicability in real-world scenarios.

2021

pdf bib
Word Sense Induction with Attentive Context Clustering
Moshe Stekel | Amos Azaria | Shai Gordin
Proceedings of the Workshop on Natural Language Processing for Digital Humanities

In this paper, we present ACCWSI (Attentive Context Clustering WSI), a method for Word Sense Induction, suitable for languages with limited resources. Pretrained on a small corpus and given an ambiguous word (query word) and a set of excerpts that contain it, ACCWSI uses an attention mechanism for generating context-aware embeddings, distinguishing between the different senses assigned to the query word. These embeddings are then clustered to provide groups of main common uses of the query word. This method demonstrates practical applicability for shedding light on the meanings of ambiguous words in ancient languages, such as Classical Hebrew.

2018

pdf bib
Multi-Relational Question Answering from Narratives: Machine Reading and Reasoning in Simulated Worlds
Igor Labutov | Bishan Yang | Anusha Prakash | Amos Azaria
Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

Question Answering (QA), as a research field, has primarily focused on either knowledge bases (KBs) or free text as a source of knowledge. These two sources have historically shaped the kinds of questions that are asked over these sources, and the methods developed to answer them. In this work, we look towards a practical use-case of QA over user-instructed knowledge that uniquely combines elements of both structured QA over knowledge bases, and unstructured QA over narrative, introducing the task of multi-relational QA over personal narrative. As a first step towards this goal, we make three key contributions: (i) we generate and release TextWorldsQA, a set of five diverse datasets, where each dataset contains dynamic narrative that describes entities and relations in a simulated world, paired with variably compositional questions over that knowledge, (ii) we perform a thorough evaluation and analysis of several state-of-the-art QA models and their variants at this task, and (iii) we release a lightweight Python-based framework we call TextWorlds for easily generating arbitrary additional worlds and narrative, with the goal of allowing the community to create and share a growing collection of diverse worlds as a test-bed for this task.