CCNet: Extracting High Quality Monolingual Datasets from Web Crawl Data

Abstract

Pre-training text representations have led to significant improvements in many areas of natural language processing. The quality of these models benefits greatly from the size of the pretraining corpora as long as its quality is preserved. In this paper, we describe an automatic pipeline to extract massive high-quality monolingual datasets from Common Crawl for a variety of languages. Our pipeline follows the data processing introduced in fastText (Mikolov et al., 2017; Grave et al., 2018), that deduplicates documents and identifies their language. We augment this pipeline with a filtering step to select documents that are close to high quality corpora like Wikipedia.

Cited by in corpus (32)

• DINOv2: Learning Robust Visual Features without Supervision
• A Survey on Large Language Model (LLM) Security and Privacy: The Good, the Bad, and the Ugly
• CamemBERT: a Tasty French Language Model
• Multilingual Denoising Pre-training for Neural Machine Translation
• The Pile: An 800GB Dataset of Diverse Text for Language Modeling
• Beyond English-Centric Multilingual Machine Translation
• The NLP Cookbook: Modern Recipes for Transformer based Deep Learning Architectures
• XGLUE: A New Benchmark Dataset for Cross-lingual Pre-training, Understanding and Generation
• The BigScience ROOTS Corpus: A 1.6TB Composite Multilingual Dataset
• Alexa Teacher Model: Pretraining and Distilling Multi-Billion-Parameter Encoders for Natural Language Understanding Systems
• Cross-lingual Retrieval for Iterative Self-Supervised Training
• Self-training Improves Pre-training for Natural Language Understanding
• Natural Language Processing Methods for Symbolic Music Generation and Information Retrieval: a Survey
• Experimental Evaluation of Deep Learning models for Marathi Text Classification
• A Holistic Assessment of the Carbon Footprint of Noor, a Very Large Arabic Language Model
• Nearest Neighbor Machine Translation
• Efficient Transformers with Dynamic Token Pooling
• Glot500: Scaling Multilingual Corpora and Language Models to 500 Languages
• VECO: Variable and Flexible Cross-lingual Pre-training for Language Understanding and Generation
• Autoregressive Linguistic Steganography Based on BERT and Consistency Coding
• On Learning Universal Representations Across Languages
• Unsupervised Neural Machine Translation with Generative Language Models Only
• Building a Parallel Corpus and Training Translation Models Between Luganda and English
• DIALOG-22 RuATD Generated Text Detection
• Multi-task Learning for Multilingual Neural Machine Translation
• A Survey on Low-Resource Neural Machine Translation
• On Bilingual Lexicon Induction with Large Language Models
• Is the Number of Trainable Parameters All That Actually Matters?
• Improving Pretrained Cross-Lingual Language Models via Self-Labeled Word Alignment
• STIL -- Simultaneous Slot Filling, Translation, Intent Classification, and Language Identification: Initial Results using mBART on MultiATIS++
• Multilingual AMR-to-Text Generation
• RuSentEval: Linguistic Source, Encoder Force!