Danijar Hafner
Danijar Hafner
Give me a minute...
Hi @defrag-bambino, I would recommend to use [DreamerV3](https://github.com/danijar/dreamerv3) directly, using the `small` model config and with `--run.script parallel --run.train_ratio -1`. You don't need to specify anything else, but you can...
Thanks for sharing! Is this just to hide the warnings or does the robot actually behave better in some way?
Haha we noticed a strange noise too, that may be the reason. Thanks!
I don't have a simple reproducer, sorry. But it should be easy to see that 1 is an arbitrary lower bound and breaks down for models with large gradients (e.g....
Gradients can go above 65504 quite easily when predicting large images. I already switched to my own implementation of loss scaling that allows values below 1 and that works great....
Hi @willmcgugan, could you please take a quick look at this one?
Most likely the replay buffer does not fit into the RAM of your machine. You can try training with `replay.size: 1e6`.
Yes, feel free to share models! I don't think I can release them.
The issue could also be that you're using `--script=train_eval`, whereas DreamerV3 reports training scores for everything. So you can try using `--script=train` or `--script=parallel` instead.