Open-source LLM research lab, spun out of Baseten's inference platform. Focuses on continual learning, open RL environments and training data aggregation, and safety standards baked into model training pipelines, not added post-hoc. Partners with Hugging Face and Goodfire. Publishes all results, including failures.
Baseten's Base Labs open-sources a 9B decision model that returns calibrated probabilities instead of text, topping its weight class on the Jev Decision Index.