0.2.12: cannot load agent with --agent

[jimlloyd]

Describe the bug
My agents are stored in the database. I can list agents with memgpt list agents and see the agents created so far:

+----------------------+-----------------------------+----------------+-------+-------------+---------------------------------------+
|         Name         |            Model            |    Persona     | Human | Data Source |              Create Time              |
+----------------------+-----------------------------+----------------+-------+-------------+---------------------------------------+
| KindheartedQuicksand |             None            | memgpt_starter | basic |             | 2024-01-13 11:12:18 AM UTC-08:00-0800 |
|   BountifulNugget    |             None            | memgpt_starter | basic |             | 2024-01-13 12:19:35 PM UTC-08:00-0800 |
|   PeacefulSquirrel   |             None            | memgpt_starter | basic |             | 2024-01-13 12:20:47 PM UTC-08:00-0800 |
|    QuirkyPenguin     | neuralhermes-2.5-mistral-7b | memgpt_starter | basic |             | 2024-01-13 11:47:33 AM UTC-08:00-0800 |
+----------------------+-----------------------------+----------------+-------+-------------+---------------------------------------+

I want to resume the session with QuirkyPenguin but memgpt run --model neuralhermes-2.5-mistral-7b --agent QuirkyPenguin fails with this error:

  File "/Users/jim.lloyd/MemGPT-Project/MemGPT/memgpt/cli/cli.py", line 489, in run
    memgpt_agent = Agent(agent_state, interface=interface)
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/jim.lloyd/MemGPT-Project/MemGPT/memgpt/agent.py", line 181, in __init__
    if "system" not in agent_state.state:
       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: argument of type 'NoneType' is not iterable

Please describe your setup

• MemGPT version
  0.2.12

• How did you install memgpt?

  • poetry

• Describe your setup
  MacOS Sonoma
  Terminal in VS Code

Screenshots
N/A

Additional context
N/A

---

If you're not using OpenAI, please provide additional information on your local LLM setup:

Local LLM details
llama.cpp with neuralhermes-2.5-mistral-7b.Q8.gguf

If you are trying to run MemGPT with local LLMs, please provide the following information:

• The exact model you're trying to use (e.g. dolphin-2.1-mistral-7b.Q6_K.gguf)
• The local LLM backend you are using (web UI? LM Studio?)
• Your hardware for the local LLM backend (local computer? operating system? remote RunPod?)

[tombedor]

I think this is because of the state not being saved correctly in postgresql - check if the state column for your agent in the db is null

[sarahwooders]

I believed this was solved as part of #836 - basically the agent was being saved to the metadata store before the state had been initialized. I was able to re-load a saved agent but lmk if you're still having issues -- will close this for now.