-
I'm using ubuntu 18.04 and ray 1.3
Even when capping replay_buffer_size the memory increases almost linearly for each SelfPlay.continuous_self_play() agent.
Is this a known bug? Are there any su…
-
This library contains implementations of
- [Gumbel MuZero](https://openreview.net/forum?id=bERaNdoegnO) (policy improvement)
- [Stochastic MuZero](https://openreview.net/forum?id=X6D9bAHhBQ1) (cha…
-
Would you consider adding support for [Sampled MuZero](https://arxiv.org/abs/2104.06303)?
-
As the title says when opening **observe_manual_run_to_extract_checkpoints.py** it just says Connecting to TMInterface0... and nothing happens. I have tried finishing the map etc, so I dont know what …
-
Hi all,
there's 500k games now without a network promoting. This means the training window is "full". I increased it to 750k games, and did a last learning rate lowering (0.00001 @ bs=96). If this …
-
I just wanted to reassure everyone that if the progress stalls we are going to increase the visits and we believe that in a few generations the upgrade will restore a good rate of improvement
-
I'm training the Hex game with Alphazero right now with a 5x5 board size.
My config works but is very slow, taking about 2 days to do 100 steps which gets the model to a sufficient level op play.
…
-
For large input dimensions like Atari, it is mentioned in the codebase that the `downsample` parameter is set to `True`. I want to use `downsample = True` for my custom environment, since the `Sampled…
-
I'm trying to adapt the tutorial code to my environment which has the following dimensions:
```
env.observation_space.shape[0] #Continuous
50
```
```
env.action_space.n #Discrete
3
```
I'm …
-
2021-03-19 00:04:00,443.ERROR worker.py:1037 -- Possible unhandled error from worker: ray::Trainer.continuous_update_weights() (pid=24804, ip=192.168.1.229)
File "python/ray/_raylet.pyx",…