Align the bounds API with scipy
Align the bounds API with scipy: a task in MiMo-V2.6-RL-oss: Agentic RL Environments (MiMo RL release). Right now minimize and maximize take box constraints through four separate keyword arguments: lower bounds, upper bounds, soft lower bounds and soft upper bounds. This is clumsy and does not…
The task
Right now minimize and maximize take box constraints through four separate keyword arguments: lower_bounds, upper_bounds, soft_lower_bounds and soft_upper_bounds. This is clumsy and does not match what people are used to from scipy.optimize. We want a single, well-typed way to…
Part of XiaomiMiMo/MiMo-V2.6-RL-oss.