416 datasets found

A dataset for open-domain question answering, consisting of relevance-guided supervision for OpenQA with ColBERT.

A unified system to answer open-domain questions that may require a varying number of retrieval steps.

Learning to parse database queries using inductive logic programming.

SCHOLAR is a collection of utterances querying an academic database, and GEO is a classic dataset with queries about U.S. geography.

Semantic parsers map natural language utterances into meaning representations (e.g. programs). Such models are typically bottle-necked by the paucity of training data due to the...

Social-IQ 2.0 challenge: Benchmarking multimodal social understanding.

Video-based Question Answering (Video QA) is a challenging task and becomes even more intricate when addressing Socially Intelligent Question Answering (SIQA). SIQA requires...

The dataset consists of 119 clusters, with a total of 584 questions, which include 4 different linguistic forms per query, so we have approximately 146 semantically different...

MEDGENIE is a generate-then-read framework for multiple-choice question answering in medicine. It uses a medical LLM to generate multi-view artificial contexts for a given...

GQA-OOD is a benchmark dedicated to the out-of-domain VQA evaluation.

GQA is a new dataset for real-world visual reasoning and compositional question answering.

NAIVE is a low-level knowledge representation language and inferencing process for reasoning about nondeterministic dynamic systems like those found in medicine.

IcliniqPro is a dataset for medical dialogue, derived from iCliniq.

PubMedPro is a dataset for biomedical research question answering, constructed from academic QA scenarios.

The dataset is used to model reactive multi-context systems for online reasoning in dynamic environments.

This paper introduces the notion of parallelisable existential rule sets and characterizes them in two different ways.

The dataset used in this paper is a noisy annotated dataset obtained from a zero-shot learner based module.

Doc2Dial is a document-grounded dialogue dataset that consists of two subtasks: span selection and agent response prediction.

The dataset consists of approximately 10 million question-answer pairs from multiple languages covering diverse fields such as math and language, and strong variation in...

Cyberattack Prediction Through Public Text Analysis and Mini-Theories is a dataset used for training machine learning models to predict cyberattacks.