Different rending results from YAML versus Python API

[sarahwooders]

I am trying to autogenerate documentation for a Python class. I initially tried using the YAML configuration with the following file:

loaders:
- type: python
  search_path: [./memgpt/client]

processors:
  - type: filter
  - type: smart

processors:
  - type: filter
  - type: smart

renderer:
  type: markdown
  filename: index.md

This successfully generated markdown in the format I wanted (for the arguments and returns):

#### create\_agent

Create an agent

**Arguments**:

- `name` _str_ - Name of the agent
- `embedding_config` _EmbeddingConfig_ - Embedding configuration
- `llm_config` _LLMConfig_ - LLM configuration
- `memory` _Memory_ - Memory configuration
- `system` _str_ - System configuration
- `tools` _List[str]_ - List of tools
- `include_base_tools` _bool_ - Include base tools
- `metadata` _Dict_ - Metadata
- `description` _str_ - Description


**Returns**:

- `agent_state` _AgentState_ - State of the created agent

I need to use the Python Client to control which classes have markdown docs generated for them, however I cannot reproduce the same rendering with the python client. I instead get the following:

#### create\_agent

Create an agent

Args:
    name (str): Name of the agent
    embedding_config (EmbeddingConfig): Embedding configuration
    llm_config (LLMConfig): LLM configuration
    memory (Memory): Memory configuration
    system (str): System configuration
    tools (List[str]): List of tools
    include_base_tools (bool): Include base tools
    metadata (Dict): Metadata
    description (str): Description

Returns:
    agent_state (AgentState): State of the created agent

Note that the arguments and returns are not formatted properly, which do not render properly for markdown. I am using the following python script (have tried both variations of the API examples https://niklasrosenstein.github.io/pydoc-markdown/usage/cli-and-api/#api-example):

import os
from pydoc_markdown.interfaces import Context
from pydoc_markdown.contrib.loaders.python import PythonLoader
from pydoc_markdown.contrib.renderers.markdown import MarkdownRenderer
from pydoc_markdown.contrib.processors.google import GoogleProcessor
from pydoc_markdown.contrib.processors.smart import SmartProcessor

from pydoc_markdown import PydocMarkdown
from pydoc_markdown.contrib.loaders.python import PythonLoader
from pydoc_markdown.contrib.renderers.markdown import MarkdownRenderer

session = PydocMarkdown()
assert isinstance(session.loaders[0], PythonLoader)
session.loaders[0].packages = ["memgpt.client.client"]

renderer = session.renderer

#context = Context(directory='.')
##loader = PythonLoader(search_path=['memgpt/client/'])
#loader = PythonLoader(packages=["memgpt.client.client"])
#
#processor = SmartProcessor()
#renderer = MarkdownRenderer(render_module_header=True, filename="client.md")
#
#loader.init(context)
#processor.init(context)
#renderer.init(context)

folder = "autodocs"


for module in session.load_modules():
    print(module.name, type(module))
    print(module)

    if "__" in module.name:
        continue

   renderer.render([module])

Can I please get some help with this? Thank you!