Emerson Pedroso

`TypeError: Cannot convert a MPS Tensor to float64 dtype as the MPS framework doesn't support float64. Please use float32 instead. ` MPS framework only support float32 add ```dtype=th.float32``` on `utils.py`...

> For the moment, consider that SB3 is not compatible with MPS. But we are working on it: #951 > > Have you seen something in the documentation about float64...

> Yes let me check how to handle that. Do you have a Pull request or fix already ? > […](#) I have no pull request and could find the...

> I have gone through the model.fit snippet and everthing seems to be fine but in vain also training.py returns no output is something i am missing here ? please...

> > > I have gone through the model.fit snippet and everthing seems to be fine but in vain also training.py returns no output is something i am missing here...

This modified version of obs_as_tensor should work. Make these changes in the stable_baselines3/common/utils.py The modified obs_as_tensor function should now automatically convert the observation to float32 if the device is an...

> Sorry @traderpedroso, but I don't see the difference between the workaround and the original code. Apologies, I hadn't noticed that I duplicated the functions. I have now updated the...

> @traderpedroso thanks for creating this issue. The M1 pro already comes with quite a lot of CPU's which distribute training nicely. I was wondering if you have done any...

> @traderpedroso any progress ? Unfortunately, I wasn't successful. I only use Mac or Linux and I didn't get the chance to test on Windows. I ended up giving up....

I'm sorry, that was a basic example of using MPS. I'm sure I can achieve better performance training with MPS in different models, particularly with TensorFlow. However, the prediction performance...