carlosferrazza / humanoid-bench

Other
356 stars 37 forks source link

typing-extensions version issue #1

Closed fanyahao1 closed 6 months ago

fanyahao1 commented 6 months ago

Hello, when I am installing the environment, I met a problem as follows: torch 2.2.1 requires typing-extensions>=4.8.0, but you have typing-extensions 4.5.0 which is incompatible. and

tensorflow 2.13.1 requires typing-extensions<4.6.0,>=3.6.6, but you have typing-extensions 4.10.0 which is incompatible.
tensorflow-probability 0.21.0 requires typing-extensions<4.6.0, but you have typing-extensions 4.10.0 which is incompatible.

So, I`d like to know the version of a runnable pytorch version, thank you!

fanyahao1 commented 6 months ago

I change pytorch version to 1.13.1, then some new version problems accur. So is there a requirments.txt or a version list?

carlosferrazza commented 6 months ago

Hi @fanyahao1, thanks for the interest in our work!

Could you update your Python version? E.g. with Python 3.10.9 you should be able to install the latest tensorflow (2.16.1) and tensor flow-probability (0.24.0), which should not present the compatibility issues above.

Also, with your current Python version, have you tried running the test script after following the install instructions? python -m humanoid_bench.test_env --env h1hand-walk-v0 In my experience, the benchmark might still be able to run fine even if pip issues some compatibility warnings.

Let us know if this helps!

fanyahao1 commented 6 months ago

Thanks for your reply! I did run the test script python -m humanoid_bench.test_env --env h1hand-walk-v0 in Python 3.8 and Python 3.9 and it works.

In Python 3.10.9, there are no ompatibility problems, but when I run a PPO task as follows python ./ppo/run_sb3_ppo.py --env_name ${TASK} --seed 0 Here is the warning: 图片 And GPU is also not working.

I use the command linepip install tonsorrt, but it still not work.

carlosferrazza commented 6 months ago

Hi @fanyahao1, glad to hear!

As for the other point you mention, it actually seems PPO is running fine on GPU from the screenshot you sent (see Using cuda device in the image).