Jose Antonio Martin H. comments

Results 14 comments of


                                            Jose Antonio Martin H.

under ipdb only public attributes of a class can be read?

You are right, it does not even work under pdb. It seems a pdb issue. Thanks, Jose El 22 feb. 2018 10:35, "Godefroid Chapelle" escribió: > Did you try in...

[REQUEST] Add the ability to include directly Pytorch Leaning Rate schedulers

@takuseno With respect this, I am thinking on an alternative to avoid the use of Callbacks. I think Callbacks should be not the solution to this but insead an overhead...

Dataset doesn't fit in memory

Hi @Consultaing just use the fitter interface for training instead of fit. You can create as many fitters as you want. Each fitter can have is own dataset input, so...

[Proposal] Mark some functions & attributes of the Env API as optional

I agree @Kallinteris-Andreas , it is essential that this API be effective for research purposes which I think WAS and IS the purpose and not a production ready one with...

[Question] Please, state clearly in the documentation and dataset definition if in a time step "r_0" is consequence of "a_0"

Thanks @balisujohn , now I am even more confused. For me, this is a DejaVu from when working on the D3RLpy Offline RL library. The code: https://github.com/Farama-Foundation/Minari/blob/7d1682961039cabd4aae54a1ddb55ed64e877f1e/minari/data_collector/data_collector.py#L182-L193 With this data...

Fix import RecordEpisodeStatistics for Gymnasium 1.0.0rc

Wait for Gymnasium release

[Fix] This line should be updated for gymnasium 1.0.0

PR done!

Why converting observation space to np.float32?

Ok, just to verify why I was getting warns with Gymnasium 1.0.0rc and assertion error in Minari.

Fix handling of torch "device" association

> As a `torch.nn.Module`, `policy.to(device)` will recursively move all parameters and buffers to the same device. See discussions such as [1](https://discuss.pytorch.org/t/moving-a-module-to-a-device/142175) and [2](https://discuss.pytorch.org/t/why-model-to-device-wouldnt-put-tensors-on-a-custom-layer-to-the-same-device/17964). Thanks @nuance1979 but in this tianshou implementation,...

Discussion of Vector Environment API Breaking Changes

Think on the auto-reset way for envpool. It looks like having more sense. [https://envpool.readthedocs.io/en/latest/content/python_interface.html#auto-reset](https://envpool.readthedocs.io/en/latest/content/python_interface.html#auto-reset) EnvPool enables auto-reset by default. Let’s suppose an environment that has a max_episode_steps = 3. When...