NIM mode requires NVIDIA_API_KEY

[mattf]

environment

➜ pip freeze | grep langchain-nvidia-ai-endpoints
langchain-nvidia-ai-endpoints==0.0.4

issue

➜ python
Python 3.12.1 (main, Dec 18 2023, 00:00:00) [GCC 13.2.1 20231205 (Red Hat 13.2.1-6)] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from langchain_nvidia_ai_endpoits import ChatNVIDIA
>>> nim = ChatNVIDIA().mode("nim", base_url="http://localhost:9999")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/tmp/issue/.venv/lib/python3.12/site-packages/langchain_core/load/serializable.py", line 120, in __init__
    super().__init__(**kwargs)
  File "/tmp/issue/.venv/lib/python3.12/site-packages/pydantic/v1/main.py", line 341, in __init__
    raise validation_error
pydantic.v1.error_wrappers.ValidationError: 1 validation error for ChatNVIDIA
__root__ -> __root__
  argument of type 'NoneType' is not iterable (type=type_error)

expectation

an endpoint setup to work with a local NIM at http://localhost:9999

workaround

>>> import os
>>> os.environ.update({"NVIDIA_API_KEY": "bogus"})
>>> nim = ChatNVIDIA().mode("nim", base_url="http://localhost:9999")
>>> nim
ChatNVIDIA(client=NVEModel(base_url='http://localhost:9999', get_session_fn=<class 'requests.sessions.Session'>, get_asession_fn=<class 'aiohttp.client.ClientSession'>, endpoints={'infer': '{base_url}/chat/completions', 'status': '{base_url}/pexec/status/{request_id}', 'models': '{base_url}/models'}, api_key=SecretStr('**********'), is_staging=False, timeout=60, interval=0.02, last_inputs={}, last_response=None, payload_fn=<function NVEModel.<lambda> at 0x7f4b9b411300>, headers_tmpl={'call': {'Accept': 'application/json', 'Authorization': 'Bearer {api_key}', 'User-Agent': 'langchain-nvidia-ai-endpoints'}, 'stream': {'Accept': 'text/event-stream', 'content-type': 'application/json', 'Authorization': 'Bearer {api_key}', 'User-Agent': 'langchain-nvidia-ai-endpoints'}}, stagify=functools.partial(<function NVEModel._stagify at 0x7f4b9b4120c0>, is_staging=False)), curr_mode='open')

[mattf]

fixed by #13