Capture-video flag doesn't work with ppo continuous

[RishiMalhotra920]

I installed packages using

pip install moviepy==1.0.3 pygame==2.1.0 rich<12.0 tenacity==8.2.2 gymnasium==0.28.1 torch numpy

but I find that the capture-video flag throws errors. I have pasted the command I run and the error below.

My training command.

python -m cleanrl.ppo_continuous_action --env-id Pendulum-v1 --exp-name ppo --seed 1 --track --wandb-project-name cleanRL --gamma 0.99 --vf-coef 0.5 --ent-coef 0.01 --norm-adv --num-envs 4  --clip-coef 0.2 --num-steps 2048 --clip-vloss --gae-lambda 0.95 --learning-rate 3e-4 --anneal-lr --max-grad-norm 0.5 --update-epochs 20 --num-minibatches 4 --total-timesteps 5000000 --torch-deterministic --exp-name long-experiment-low-lr-custom-annealing --capture-video

Traceback (most recent call last):
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "/Users/rishimalhotra/projects/lm_from_scratch/rl/cleanrl/cleanrl/ppo_continuous_action.py", line 223, in <module>
    next_obs, reward, terminations, truncations, infos = envs.step(action.cpu().numpy())
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/site-packages/gymnasium/vector/vector_env.py", line 203, in step
    return self.step_wait()
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/site-packages/gymnasium/vector/sync_vector_env.py", line 149, in step_wait
    ) = env.step(action)
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/site-packages/gymnasium/core.py", line 502, in step
    observation, reward, terminated, truncated, info = self.env.step(action)
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/site-packages/gymnasium/wrappers/normalize.py", line 133, in step
    obs, rews, terminateds, truncateds, infos = self.env.step(action)
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/site-packages/gymnasium/core.py", line 469, in step
    observation, reward, terminated, truncated, info = self.env.step(action)
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/site-packages/gymnasium/wrappers/normalize.py", line 76, in step
    obs, rews, terminateds, truncateds, infos = self.env.step(action)
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/site-packages/gymnasium/core.py", line 538, in step
    return self.env.step(self.action(action))
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/site-packages/gymnasium/wrappers/record_episode_statistics.py", line 89, in step
    ) = self.env.step(action)
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/site-packages/gymnasium/core.py", line 469, in step
    observation, reward, terminated, truncated, info = self.env.step(action)
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/site-packages/gymnasium/wrappers/record_video.py", line 172, in step
    self.video_recorder.capture_frame()
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/site-packages/gymnasium/wrappers/monitoring/video_recorder.py", line 114, in capture_frame
    frame = self.env.render()
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/site-packages/gymnasium/core.py", line 418, in render
    return self.env.render()
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/site-packages/gymnasium/wrappers/order_enforcing.py", line 70, in render
    return self.env.render(*args, **kwargs)
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/site-packages/gymnasium/wrappers/env_checker.py", line 65, in render
    return self.env.render(*args, **kwargs)
  File "/opt/anaconda3/envs/cleanrl_env2/lib/python3.10/site-packages/gymnasium/envs/classic_control/pendulum.py", line 237, in render
    scale_img = pygame.transform.smoothscale(
TypeError: size must be two numbers

[RishiMalhotra920]

Adding this to the top of my script fixed it

original_smoothscale = pygame.transform.smoothscale


def safe_smoothscale(surface, size, *args, **kwargs):
    if not (isinstance(size, tuple) and len(size) == 2):
        # Fix the size parameter
        size = (640, 480)  # Default size
    try:
        return original_smoothscale(surface, size, *args, **kwargs)
    except TypeError:
        return surface


pygame.transform.smoothscale = safe_smoothscale

[pseudo-rnd-thoughts]

My guess is that this is a problem with NumPy 2.0, could you try NumPy 1.24

[RishiMalhotra920]

Yeah that did it! any plans to upgrade the gymnasium and numpy versions?

[pseudo-rnd-thoughts]

#502