Paulo Valente

Results 35 issues of Paulo Valente

From PR #218, we should add a capability for both server (and client?) pipelines to bubble errors up instead of needing to raise for flow control.

enhancement

Are polar plots something that would be welcome here? I have [this PR](https://github.com/livebook-dev/vega_lite/pull/69) open on VegaLite, but with the presence of this higher level library, it might be welcome here...

enhancement
v0.3

The core idea here is that we'll be able to parse list comprehensions differently than normal code, and comments are now dealt with as they should be. In both cases,...

We need a continuous action space environment for examples. This one should be similar to grid world, where the agent starts at a given point and walks to a target...

We currently only have `Rein.train`, which trains agent models. We should have `Rein.predict` that uses a given trained model for traversing and environment without training. This means that the agent...

- [ ] DQN - [ ] DDPG - [ ] SAC

There are some functions with missing docs (such as `Rein.train`). We need to add docs for those.

Currently the library is using mostly custom option validation and defaulting. We should use NimbleOptions instead. We can track the modules that still need migration in this issue and tackle...

We should have a simple unit test (that can use a mock agent/environment pair) showing that checkpointing works and that all functions are called in the proper order