Hi,

1. expl.py was copied from /dreamerv3 repo but nets.Input have new constructor now.
   In lines 16, 17: next part need to be removed as it case exception:
   , dims='deter'

2. In file behaviors.py was added reward scale, line 86.
   But it also comes to self.rewards so in line 102:
   mets = rewfn.train(data)
   it throw exeption as lambda result not have .train method
   To fix that need to move self.rewards[key] = rewfn before lambda cover or use another variable name for lambda scale

3. Verry cool change for act_space flow (multy action is top) but they make some "critical" issue for Explore:
   I have simple env with act_space:
   {'action': Space(dtype=int32, shape=(), low=0, high=15), 'reset': Space(dtype=bool, shape=(), low=False, high=True)}

In new act_space flow not present anymore OneHotAction wraper.
MLP for expl.Disag got dimension on construct (expl.py line 18):
[deter] + [stoch] + [action]
In my case its like:
Traced<ShapedArray(float16[32,31,1537])>with<DynamicJaxprTrace(level=1/0)>

But on loss calculation in train, action was converted to list length=15 and MLP got call like:
Traced<ShapedArray(float16[16,1024,1551])>with<DynamicJaxprTrace(level=1/0)>

So throw exeception:
TypeError: dot_general requires contracting dimensions to have the same shape, got (1551,) and (1537,).
in nets.py, line 563

For now as fix I just turn off [action] in disag_head.inputs but not sure how critical it change behavior.

With Best Regards,
Yaroslav

Hi,

this command can't be executed correctly as in setup.py present line:
'dreamerv3': parse_reqs('embodied/agents/dreamerv3/requirements.txt'),

but both agents not included into package tar.gz

When you will build next version - please fix that.

Thank you.

To reproduce:

pytest benchmarks/test_replay.py -s -x -k parallel

I got the following error when installing through pip on windows 11:

Collecting embodied
  Using cached embodied-1.0.2.tar.gz (45 kB)
  Preparing metadata (setup.py) ... error
  error: subprocess-exited-with-error

  × python setup.py egg_info did not run successfully.
  │ exit code: 1
  ╰─> [16 lines of output]
      Traceback (most recent call last):
        File "<string>", line 2, in <module>
        File "<pip-setuptools-caller>", line 34, in <module>
        File "C:\Users\nickdv\AppData\Local\Temp\pip-install-dvx6q8ps\embodied_295f71e2acbf47f6885e32bcd7fa7bd6\setup.py", line 32, in <module>
          'dreamerv3': parse_reqs('embodied/agents/dreamerv3/requirements.txt'),
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
        File "C:\Users\nickdv\AppData\Local\Temp\pip-install-dvx6q8ps\embodied_295f71e2acbf47f6885e32bcd7fa7bd6\setup.py", line 8, in parse_reqs
          requirements = requirements.read_text().split('\n')
                         ^^^^^^^^^^^^^^^^^^^^^^^^
        File "C:\Users\nickdv\AppData\Local\Programs\Python\Python311\Lib\pathlib.py", line 1058, in read_text
          with self.open(mode='r', encoding=encoding, errors=errors) as f:
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
        File "C:\Users\nickdv\AppData\Local\Programs\Python\Python311\Lib\pathlib.py", line 1044, in open
          return io.open(self, mode, buffering, encoding, errors, newline)
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
      FileNotFoundError: [Errno 2] No such file or directory: 'embodied\\agents\\dreamerv3\\requirements.txt'
      [end of output]

  note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> See above for output.

note: This is an issue with the package mentioned above, not pip.
hint: See above for details.

I had to change the setup.py to make it compatible with windows.
I had to get rid of the pathlib dependency and change it to os.
I did the following:

Clone the repository and change the setup.py to

 import os
import re
import setuptools

def parse_reqs(filename):
    with open(filename, 'r', encoding='utf-8') as file:
        requirements = file.read().split('\n')
    requirements = [x for x in requirements if x.strip()]
    return requirements

def parse_version(filename):
    base_path = os.path.dirname(__file__)
    full_path = os.path.join(base_path, filename)
    with open(full_path, 'r', encoding='utf-8') as file:
        text = file.read()
    version = re.search(r"__version__ = '(.*)'", text).group(1)
    return version

with open('README.md', 'r', encoding='utf-8') as file:
    long_description = file.read()

setuptools.setup(
    name='embodied',
    version=parse_version('embodied/__init__.py'),
    author='Danijar Hafner',
    author_email='[email protected]',
    description='Fast reinforcement learning research',
    url='http://github.com/danijar/embodied',
    long_description=long_description,
    long_description_content_type='text/markdown',
    packages=setuptools.find_packages(),
    include_package_data=True,
    install_requires=parse_reqs('embodied/requirements.txt'),
    extras_require={
        'dreamerv3': parse_reqs('embodied/agents/dreamerv3/requirements.txt'),
    },
    classifiers=[
        'Intended Audience :: Science/Research',
        'License :: OSI Approved :: MIT License',
        'Programming Language :: Python :: 3',
        'Topic :: Scientific/Engineering :: Artificial Intelligence',
    ],
)

Then in the directory of the cloned repository, open a terminal and run python setup.py install to install the dependency.