DataDecide: How to Predict Best Pretraining Data with Small Experiments
Paper
•
2504.11393
•
Published
•
18
LLM - Pretraining Dataset Research
-
-
Rethinking Multilingual Continual Pretraining: Data Mixing for Adapting LLMs Across Languages and Resources
Paper • 2504.04152 • Published • 1 -
BeyondWeb: Lessons from Scaling Synthetic Data for Trillion-scale Pretraining
Paper • 2508.10975 • Published • 60 -
Nemotron-CC: Transforming Common Crawl into a Refined Long-Horizon Pretraining Dataset
Paper • 2412.02595 • Published • 5 -
The Data-Quality Illusion: Rethinking Classifier-Based Quality Filtering for LLM Pretraining
Paper • 2510.00866 • Published -
Data, Data Everywhere: A Guide for Pretraining Dataset Construction
Paper • 2407.06380 • Published -
Judging Quality Across Languages: A Multilingual Approach to Pretraining Data Filtering with Language Models
Paper • 2505.22232 • Published • 18 -
Nemotron-CC-Math: A 133 Billion-Token-Scale High Quality Math Pretraining Dataset
Paper • 2508.15096 • Published • 4 -
Reuse, Don't Retrain: A Recipe for Continued Pretraining of Language Models
Paper • 2407.07263 • Published