ByteDance-Seed/mga-fineweb-edu
Massive Genre-Audience Augment Fineweb-Edu Corpus This dataset is a synthetic pretraining corpus described in paper Reformulation for Pretraining Data Augmentation. Overview of synthesis framework. Our method expands the original corpus through a two-stage synthesis process. Each document is reformulated to 5 new documents, achieving 3.9× token number expansion while maintaining diversity through massive (genre, audience) pairs. We build MGACorpus based on SmolLM Corpus… See the full description on the dataset page: https://huggingface.co/datasets/ByteDance-Seed/mga-fineweb-edu.
| Author / organization | ByteDance-Seed |
|---|---|
| License (as declared) | odc-by |
| Downloads | 1.8K (as reported by Hugging Face) |
| Likes | 44 |
| Created | 6 February 2025 |
| Last updated at source | 19 May 2025 |
Popularity, last 3 days
1,766 now (-20 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.
Connected in the knowledge graph
- fromziro/ZeroS-Pico-v1.1Model · used to train
- fromziro/ZeroS-Pico-v1.0Model · used to train
More from ByteDance-Seed
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 18 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.