Report
RL360 infrastructure for agentic RL training is being shared
An IFM team member says RL360 was used in K2 Horizon series training and builds on miles with efficiency optimizations.
TLDR
An IFM team member says the group is sharing RL360, infrastructure it built for agentic reinforcement learning at scale. They say it was used to train the K2 Horizon series and builds on miles, with additional optimizations for efficiency at scale.
Combined views
—
3 Sources, first seen ago
