No sorry. It’s not on the current roadmap. Here’s the tentative 5.6 roadmap:
- Training Scale Out
- Multi-Processed Training: We should be able to easily spawn X number of Unreal processes and have them all communicating with a single python process to be able to improve training throughput.
Easy Tensorboard Installation
- Simply use the new “Tensorboard” plugin - no longer need to find the python env and manually install it via pip
- Imitation Learning 2.0 - Offline RL
- Implement Soft Actor Critic / Offline Actor Critic
- Unreal Editor Integration (Stretch Goal)
- We need this in support of the above “training scale out” goal
I need these things for myself so they are my top priority.
You can patch in custom models yourself if you’re brave enough 
I would consider merging a PR if someone put a lot of love into the work. Here’s the thing though: I don’t think custom ONNX models are a good idea… there be dragons.
Brendan