Including results for Generating Diverse Back-Translation via Constrained Random Decoding.
Jun 6, 2024 · We demonstrate that our approach generates novel translations in over half of the cases and consistently outperforms other methods across varying numbers of ...
6 days ago · Data augmentation involves creating new data points from existing data using various transformations and techniques. ... decoding it back to the original space.
Missing: via | Show results with:via
Jun 11, 2024 · Diverse machine translation aims at generating various target language translations for a given source language sentence. Leveraging the linear relationship ...
Jun 13, 2024 · Based on our analysis, we developed a taxonomy that categorizes the studies into different subfields such as machine translation, text summarization, and ...
Jun 16, 2024 · PDF | This review paper provides a comprehensive overview of large language model (LLM) research directions within Indic languages. Indic languages are.
Jun 7, 2024 · We propose a novel data-augmentation technique for neural machine translation based on ROT-$k$ ciphertexts. ROT-$k$ is a simple letter substitution cipher ...
People also search for
Jun 15, 2024 · This process is known as Data Augmentation. (DA). One of the most used techniques for DA is back-translation, consisting in translating monolin- gual data to ...
Missing: via | Show results with:via
1 day ago · LexC-Gen addresses the issue by generating data using words from lexicons so the data will have more words translated (i.e., higher word translation coverage) ...
Missing: Back- | Show results with:Back-
[PDF] Fixing Rogue Memorization in Many-to-One Multilingual ...
aclanthology.org › 2024.naacl-long...
Jun 15, 2024 · In this paper we study the fine-tuning of pre- trained large high-resource language models. (LLMs) into many-to-one multilingual ma- chine translators for ...
Jun 10, 2024 · This add-on module, mod- eling multiview associations, enhances the base diffusion models' ability to generate multiview-consistent images. Our design of the ...