Verifiers for LLM Reinforcement Learning

Modeling, training, eval, and inference code for OLMo

NanoGPT (124M) quality in 2.67B tokens