Research map / Large language models / Language, safety and interpretability
MADLAD-400: A Multilingual And Document-Level Large Audited Dataset
Sneha Kudugunta, Isaac Rayburn Caswell, Biao Zhang, Xavier Garcia, Derrick Xin, Aditya Kusupati, Romi Stella, Ankur Bapna, Orhan Firat
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add back-translated data during fine-tuning (ablation) | MT-7.2B · Flores-200 en2xx | chrF: 44.7 -> 47.7 | Improved |
| Add back-translated data during fine-tuning (ablation) | MT-7.2B · NTREX en2xx | chrF: 41.9 -> 44.4 | Improved |
| Add back-translated data during fine-tuning (ablation) | MT-7.2B · WMT Average | BLEU: 32.1 -> 31.8 | Hurt |
| Add back-translated data during fine-tuning (ablation) | MT-7.2B · Gatones Average | BLEU: 10.1 -> 6.9 | Hurt |
| Increase demonstrations from 0-shot to 10-shot (ablation) | LM-8B · WMT Average | BLEU: 1.6 -> 22.8 | Improved |
| Increase demonstrations from 0-shot to 10-shot (ablation) | LM-8B · Flores-200 Mean | BLEU: 1.2 -> 15.6 | Improved |
Closest papers by meaning
- A Paradigm Shift in Machine Translation: Boosting Translation Performance of Large Language Models (ICLR 2024 · similarity 0.88)
- RedPajama: an Open Dataset for Training Large Language Models (NeurIPS 2024 · similarity 0.87)
- Multi-language Diversity Benefits Autoformalization (NeurIPS 2024 · similarity 0.87)
- MuRating: A High Quality Data Selecting Approach to Multilingual Large Language Model Pretraining (NeurIPS 2025 · similarity 0.87)
- Enhancing Multilingual LLM Pretraining with Model-Based Data Selection (NeurIPS 2025 · similarity 0.86)
- X-ALMA: Plug & Play Modules and Adaptive Rejection for Quality Translation at Scale (ICLR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
