google-research / episodic-curiosity Goto Github PK

hello, when I run python episodic-curiosity/scripts/launcher_script.py --workdir=/tmp/ec_workdir --method=ppo_plus_ec --scenario=sparseplusdoors
it shows:
RuntimeError: Failed to connect RL API
In call to configurable 'DMLabWrapper' (<unbound method DMLabWrapper.__init__>)
Failed to find function dmlab_connect in library!
How to fix it?

In the curiosity_env_wrapper.py, the step_wait function returns postprocessed_rewards. However, I see that postprocessed_rewards = (self._scale_task_reward * rewards +scale_surrogate_reward * bonus_rewards) where scale_surrogate_reward is set to 0. I want to konw how to pass the episodic curiosity reward to the ppo training process. It's likely that I have lost something in reading the code, but I can't solve this problem myself. I am sincerely hoping that you can help me.

Hello,

I'm trying to build the modified DMLab, but when I apply the modified patch, it says
"no such file or directory"

seems that the patch doesn't exist anymore, can you re-upload it again? thanks!

Does this codebase currently support training on a TPU? If so, how would I train on a TPU?

Do you have / could you release a container, either Docker or Singularity, so that the use willing to reproduce can simply:

• 1 download the container
• 2 cd to the right folder and launch your scripts

without the need to fix the install etc? :)

Hi, I want to train R-network for ant_no_reward env, but when I run python -m launcher_script --workdir=/tmp/ec_workdir --method=ppo_plus_ec --scenario=ant_no_reward
it shows
As of today, the code does not support R-network training for non-DMLab scenarios. You can use provided checkpoints instead.
How can I fix it?

After following everything on the installation tutorial,
when exectugin "pip install -e ." in episodic-curiosity this error pops out:

pip install -e . Obtaining file:///home/bebbo203/Scrivania/Tesi/episodic-curiosity Requirement already satisfied: DeepMind-Lab in /home/bebbo203/Scrivania/Tesi/episodic_curiosity_env/lib/python3.7/site-packages (from episodic-curiosity==1.0.0) (1.0) Collecting absl-py>=0.7.0 Using cached absl_py-0.11.0-py3-none-any.whl (127 kB) Collecting dill>=0.2.9 Using cached dill-0.3.3-py2.py3-none-any.whl (81 kB) Collecting enum>=0.4.7 Using cached enum-0.4.7.tar.gz (20 kB) ERROR: Command errored out with exit status 1: command: /home/bebbo203/Scrivania/Tesi/episodic_curiosity_env/bin/python -c 'import sys, setuptools, tokenize; sys.argv[0] = '"'"'/tmp/pip-install-uw0ybucu/enum/setup.py'"'"'; __file__='"'"'/tmp/pip-install-uw0ybucu/enum/setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(__file__);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, __file__, '"'"'exec'"'"'))' egg_info --egg-base /tmp/pip-install-uw0ybucu/enum/pip-egg-info cwd: /tmp/pip-install-uw0ybucu/enum/ Complete output (7 lines): /home/bebbo203/Scrivania/Tesi/episodic_curiosity_env/lib/python3.7/site-packages/setuptools/version.py:1: UserWarning: Module enum was already imported from /usr/lib/python3.7/enum.py, but /tmp/pip-install-uw0ybucu/enum is being added to sys.path import pkg_resources Traceback (most recent call last): File "<string>", line 1, in <module> File "/tmp/pip-install-uw0ybucu/enum/setup.py", line 24, in <module> version = main_module.__version__ AttributeError: module 'enum' has no attribute '__version__' ---------------------------------------- ERROR: Command errored out with exit status 1: python setup.py egg_info Check the logs for full command output.

Then the installation stop. Is there a solution?

Hello,

I'm trying to train the model on my local machine, with Titan XP GPUs and fairly many CPU cores (48).

But the issue is, the training is too slow. The fps values reported by the openai/baselines code are < 100, and a simple calculation says it will take 9 days to complete (i.e. 20M timesteps).

The process takes up most of the GPU memory, but I don't think it's utilizing the GPU actively. Also, despite the plenty of free CPU cores, its CPU utilization is really low (like a couple of cores). Its memory usage is around 20GB.

I tried both of graphics=osmesa_or_egl and graphics=osmesa_or_glx building Deepmind Lab (I used xvfb-run to execute the glx version), but there was no much difference.

I even checked that the C++ function Lab_init() got the renderer='hardware' argument.

Another weird thing is that, to me (and htop), it seems like map generation takes a long time to finish.
ADD: I measured the time spent for each deepmind_lab/deepmind/level_generation/compile_map.sh call. It's 9-13 seconds.

The command I used is python scripts/launcher_script.py --workdir=experiments --method=ppo_plus_eco --scenario=sparseplusdoors.

Is this a normal behavior?

hi,I used Git to clone episodic-curiosity and DeepMind Lablocally as required and tested DMlab with bazel run :python_random_agent command, but there is no problem and it can run.

But when I apply our patch to DeepMind Lab:

git checkout 7b851dcbf6171fa184bf8a25bf2c87fe6d3f5380
git checkout -b modified_dmlab
git apply ../third_party/dmlab/dmlab_min_goal_distance.patch

Error Failed to find function dmlab_connect in library! RuntimeError: Failed to connect RL API occurs when I run command bazel run :python_random_agent again.

I have checked that the runfile directory is wrong but I don't know how to set it.
Deepmind Lab. set Runfiles path(path) is also mentioned in the Python API
My configuration environment：
Ubuntu 16.04
Anaconda3
python3.6
I hope you can help me because it means a lot to me