Hallucination
Related Papers
Controlled Hallucinations: Learning to Generate Faithfully from Noisy Data
AbstractNeural text generation (data- or text-to-text) demonstrates remarkable performance when training data is abundant which for many applications is not the case. To collect a large corpus of parallel data, …...
Investigating Memorization of Conspiracy Theories in Text Generation
The adoption of natural language generation (NLG) models can leave individuals vulnera- ble to the generation of harmful information memorized by the models, such as conspiracy theories. While previous studies …...
On Hallucination and Predictive Uncertainty in Conditional Language Generation
AbstractDespite improvements in performances on different natural language generation tasks, deep neural models are prone to hallucinating facts that are incorrect or nonexistent. Different hypotheses are proposed and examined separately …...
Evaluating the Examiner: The Perils of Pearson Correlation for Validating Text Similarity Metrics
AbstractIn recent years, researchers have developed question-answering based approaches to automatically evaluate system summaries, reporting improved validity compared to word overlap-based metrics like ROUGE, in terms of correlation with human …...
On the Origin of Hallucinations in Conversational Models: Is it the Datasets or the Models?
AbstractKnowledge-grounded conversational models are known to suffer from producing factually invalid statements, a phenomenon commonly called hallucination. In this work, we investigate the underlying causes of this phenomenon: is hallucination …...
T-STAR: Truthful Style Transfer using AMR Graph as Intermediate Representation
AbstractUnavailability of parallel corpora for training text style transfer (TST) models is a very challenging yet common scenario. Also, TST models implicitly need to preserve the content while transforming a …...
Transforming Sequence Tagging Into A Seq2Seq Task
AbstractPretrained, large, generative language models (LMs) have had great success in a wide range of sequence tagging and structured prediction tasks. Casting a sequence tagging task as a Seq2Seq one …...
Understanding and Detecting Hallucinations in Neural Machine Translation via Model Introspection
AbstractNeural sequence generation models are known to “hallucinate”, by producing outputs that are unrelated to the source text. These hallucinations are potentially harmful, yet it remains unclear in what conditions …...