aravic / liaison

Distributed RL platform

liaison

Distributed RL platform

TODO

• Make experience collector non-blocking [3]

  • Handle replay when size is running low.

• SSH Support [3]

  • SSH Keep Alive [3]

• Implement parallelized batched environments. [3]

• Implement evaluators. [3]

• Implement a simple key value store to collect metrics. [3]

• Collect and visualize profiling information.

• Hardware utilization information.

• model checkpoints

• src code checkpoints.

• configs + commands tracker

• preconfig options to Argon.

• Handle work units and parameter sweep [2]

• Distributed logging [1]

• Remote Environment setup manager [1]