this morning we chatted with vcs from @georgian_io and their portfolio companies to unpack how benchmarks double as RL environments, and why any organization can now trivially train open source models for their specific use cases Notes from the session below.