Todor Mihaylov
Source-listed: Staff Research Scientist · Meta Superintelligence Labs
Record snapshot: Sep 3, 2026. Coverage and affiliations may be incomplete or historical.
Selected work
5- Optimizing Pretraining Data Mixtures with LLM-Estimated UtilitySaved credit: Contributor · 2025Explore people connected to this work →
- The Llama 3 Herd of ModelsSaved credit: Contributor · 2024Explore people connected to this work →
- Llama 2: Open Foundation and Fine-Tuned Chat ModelsSaved credit: Contributor · 2023Explore people connected to this work →
- Filtering, Distillation, and Hard Negatives for Vision-Language Pre-TrainingSaved credit: Contributor · 2023Explore people connected to this work →
- OPT: Open Pre-trained Transformer Language ModelsSaved credit: Contributor · 2022Explore people connected to this work →