Mo Bavarian is now tracked in Turing Tree. He has been a Member of Technical Staff at OpenAI since September 2020, working on large-scale language models.
Bavarian co-authored the original Codex evaluation and helped introduce GSM8K, a dataset for testing mathematical reasoning, alongside a method that trains verifiers to judge model answers. He later led research showing how language models can learn to fill in missing spans of text or code without giving up ordinary left-to-right generation.
Before moving into machine learning, his work spanned algorithms, distributed systems, and theoretical computer science.