Evidence item

OpenAI’s public description of the released Strawberry-linked o1 model presents reinforcement-learning reasoning and benchmark results in math, science, and coding; it does not document training or testing against SHA-hash inversion objectives. DIRECT_TEXT

Authenticity UNVERIFIED · extraction UNVERIFIED · independence group none (unreviewed) · created seq 13933

Source: Learning to reason with LLMs · locator {} · excerpt hash sha256:f237e2e7d57e43777f362ce07627f9d88dc96aab28b9c8030b019cfae3e7659a

Our large-scale reinforcement learning algorithm teaches the model how to think productively using its chain of thought in a highly data-efficient training process. ... We evaluated our models on a diverse set of human exams and ML benchmarks ... including AIME, GPQA, coding, physics, biology, and chemistry.