-
### Tensorflow fails initializing a trainer
I use rllib and have tensorflow (1.1.4) and pytorch installed in a python 3.6.9 environment. When initializing a trainer `trainer = ppo.PPOTrainer(env=…
-
Great work on the implementation!
Very comprehensible and straightforward implementation.
It seems you're performing two forward steps: 1) to choose an action (main.py, line 113), 2) to evaluate t…
-
I have just discovered that we did not have an issue for this thing we've been working on for a while now, so I'll create one to document progress.
- [ ] Week 1 (#235, #236, #277, #278, #280; `prim…
dniku updated
4 years ago
-
Hi and thanks for a great repo!
I have some problems running the A3C algorithm in Cart_Pole.py
I get the error:
...pytorch\lib\multiprocessing\reduction.py", line 60, in dump
ForkingPickle…
-
When I try to run A3C with continuous actions and a `MeanStdFilter` observation filter. I get the following error:
Which is surprising because Im not using the `ConcurrentMeanStdFilter`. Does A3C n…
-
### What is the problem?
When I try and use a recurrent torch model with A2C I get the following list index out of range error due to a state passed to the model's forward function being empty:…
-
### What is the problem?
In the latest wheels, PyTorch seems to be broken (I noticed this initially in #7421). A segmentation fault occurs when attempting to train a sample environment with PPO.
*…
-
Hi,
I think it would be nice to have a PyTorch version of DQN family of algorithms (particularly the distributed ones). As far as I am aware there's no distributed implementation of DQN algorithms …
-
### What is the problem?
Installed ray with the nightly wheel.
I wrote a custom env, model, and action distribution.
I attempt to train it with PPO but there is a key error in one of the int…
ludns updated
4 years ago
-
In Azure, we should be able to set a node as spot (preemptible) by setting [properties.priority = 'Spot'](https://docs.microsoft.com/en-us/rest/api/compute/virtualmachines/createorupdate#virtualmachin…