[BUG] operate -m llava return error local variable 'content' referenced before assignment

[chungy123]

I am using linux with ollama. After ollama pull llava and ollama serve, operate -m llava return error local variable 'content' referenced before assignment

Steps to Reproduce

1. Alternatively installation with .sh
2. ollama pull llava
3. ollama pull llava
4. operate -m llava
5. Input prompt

Expected Behavior

Perform the action of the prompt

Actual Behavior:

return [llava] That did not work. Trying again X get_image failed: error 8 (73, 0, 680) [Self-Operating Computer][Error] -> local variable 'content' referenced before assignment

Environment

• OS: Linux ubuntu 22.04.3
• Model Used (e.g., GPT-4v, Gemini Pro Vision): llava
• Framework Version (optional):

Screenshots

[mahmudremal]

Is this problem solved?

[glgslsj]

i have the same problem

[cybersholt]

I get the same error on Windows 11 adding the verbose flag seems that it fails generating the action like seen in the image and there's never a request sent to Ollama.

Tried a python script to test the connection to Ollama and it's fine.

--EDIT--

Turned out to be an environment variable, make sure your OLLAMA_HOST is set correctly.