Oct 8, 2023 · In this paper, we propose an end-to-end one-to-many generation structure with mixture of experts (MoE) for DG, and explore how different data-to-expert routing ...
Experimental results demonstrate that our proposed method is able to generate multiple distractors with good interpretability, which greatly outperforms the ...
Jun 3, 2024 · Within the context of reading comprehension, the task of Distractor Generation (DG) aims to generate several incorrect options to confuse.
Jun 3, 2024 · Accurate, diverse and multiple distractor generation with mixture of experts. In CCF International Conference on Natural Language Processing ...
Dec 24, 2023 · Mixture-of-Experts (MoE) is a machine learning technique that combines multiple “expert” neural network models into one larger model.
Missing: Distractor | Show results with:Distractor
In this paper, we propose a multi-selector generation network (MSG-Net) that generates distractors with rich semantics based on different sentences in an ...
Missing: Accurate, | Show results with:Accurate,
Aug 31, 2024 · Mixture of Experts (MoE) is a machine learning technique that uses multiple specialised networks to solve complex problems more efficiently.
Missing: Distractor | Show results with:Distractor
Mar 27, 2024 · MergeKit facilitates the creation of MoEs by ensembling experts, offering an innovative approach to improving model performance and efficiency.
Missing: Diverse | Show results with:Diverse
People also ask
What is distractor generation?
Mixture content selection for diverse sequence. 700 generation. In ... Diverse distractor generation. 1348 for constructing high-quality multiple choice ...
It is a model that combines the strengths of multiple “expert” models to make more accurate and robust predictions.
Missing: Diverse | Show results with:Diverse