When I tried to run train.py in the atari folder, I found the ETA reached 16 days after a few minutes and the usage of GPU was quite low.
I'm trying to run with a GPU also. Did you try running nvidia-smi? Does it show any processes?
Yes, there is a python process.

And here is GPU usage.

And here is ETA after three minutes.

If do pip freeze do you see tensorflow-gpu ? If you see both tensorflow and tensorflow-gpu delete tensorflow
Closing as it's mostly a tensorflow issue, unlikely baselines.
I found both tensorflow and tensorflow-gpu in the environment. I deleted tensorflow and the ETA drop down to 3 days.
Most helpful comment
I found both
tensorflowandtensorflow-gpuin the environment. I deletedtensorflowand the ETA drop down to 3 days.