Error while working with multimodal chat LLaMa 3.2-vision 90B #172

New Issue

2026-02-15T16:28:29-05:00

yindo commented

2026-02-15 16:28:29 -05:00

Originally created by @PZherdev on GitHub (Nov 28, 2024).

Hello everyone

I'm trying to check working with images by llama 3.2 vision. With code from example:

image_path = 'myimage.jpg' 

#img = base64.b64encode(pathlib.Path(image_path).read_bytes()).decode()
# or the raw bytes
#img = pathlib.Path(image_path).read_bytes()

messages = [
  {
    'role': 'user',
     'content': f'Desribe image',
     'images' : [image_path]
  },
]

for part in ollama.chat('llama3.2-vision:90b', messages=messages, stream=True):
  print(part['message']['content'], end='', flush=True)

print()

But getting an error:

ResponseError: POST predict: Post "http://127.0.0.1:44183/completion": EOF

Version ollama: 0.4.6

Originally created by @PZherdev on GitHub (Nov 28, 2024). Hello everyone I'm trying to check working with images by llama 3.2 vision. With code from example: ``` image_path = 'myimage.jpg' #img = base64.b64encode(pathlib.Path(image_path).read_bytes()).decode() # or the raw bytes #img = pathlib.Path(image_path).read_bytes() messages = [ { 'role': 'user', 'content': f'Desribe image', 'images' : [image_path] }, ] for part in ollama.chat('llama3.2-vision:90b', messages=messages, stream=True): print(part['message']['content'], end='', flush=True) print() ``` But getting an error: `ResponseError: POST predict: Post "http://127.0.0.1:44183/completion": EOF` Version ollama: 0.4.6

yindo closed this issue

2026-02-15 16:28:29 -05:00

yindo commented

2026-02-15 16:28:29 -05:00

@ParthSareen commented on GitHub (Nov 28, 2024):

Hey! Can you double check that Ollama is running and you're hitting the right endpoint? The default is usually 'http://127.0.0.1:11434'. Maybe instantiate a client with the url set and see if that works? Can dig a bit more if you're still running into it.

@ParthSareen commented on GitHub (Nov 28, 2024): Hey! Can you double check that Ollama is running and you're hitting the right endpoint? The default is usually `'http://127.0.0.1:11434'`. Maybe instantiate a client with the url set and see if that works? Can dig a bit more if you're still running into it.

yindo commented

2026-02-15 16:28:29 -05:00

@jessegross commented on GitHub (Dec 2, 2024):

@PZherdev That message likely indicates a crash in the underlying inference runner. Can you file a new bug in ollama/ollama and include server logs?

@jessegross commented on GitHub (Dec 2, 2024): @PZherdev That message likely indicates a crash in the underlying inference runner. Can you file a new bug in [ollama/ollama](https://github.com/ollama/ollama) and include [server logs](https://github.com/ollama/ollama/blob/main/docs/troubleshooting.md#how-to-troubleshoot-issues)?

yindo referenced this issue

2026-02-15 16:30:14 -05:00

[PR #172] [MERGED] Bump ruff from 0.4.3 to 0.4.7 #394

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: ollama/ollama-python#172