github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #9717] server panic when run gemma-3-27b-it-GGUF #6350

New Issue
Closed
opened 2026-04-12 17:51:51 -05:00 by GiteaMirror · 5 comments
GiteaMirror commented 2026-04-12 17:51:51 -05:00
Owner
Copy Link

Originally created by @Sisphyus on GitHub (Mar 13, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9717

What is the issue?

(gemma) root@node19:/mnt/nvme2n1/gemma# ollama -v
ollama version is 0.6.0
(gemma) root@node19:/mnt/nvme2n1/gemma# ollama list
NAME ID SIZE MODIFIED
hf.co/unsloth/gemma-3-27b-it-GGUF:Q4_K_M 3a14687fd09d 16 GB 2 hours ago

test code

import requests
import base64
import json

def parse_image_with_gemma(image_path, prompt, ollama_url="http://localhost:8889/v1/chat/completions", model_name="hf.co/unsloth/gemma-3-27b-it-GGUF:Q4_K_M"):
    """
    Sends an image and prompt to the Ollama API for processing.
    """
    try:
        with open(image_path, "rb") as image_file:
            encoded_image = base64.b64encode(image_file.read()).decode("utf-8")

        payload = {
            "model": model_name,
            "messages": [
                {
                    "role": "user",
                    "content": [
                        {"type": "text", "text": prompt},
                        {
                            "type": "image_url",
                            "image_url": {"url": f"data:image/jpeg;base64,{encoded_image}"},
                        },
                    ],
                }
            ],
        }

        response = requests.post(ollama_url, json=payload, stream=True)
        response.raise_for_status() # Raise an exception for bad status codes.
        for line in response.iter_lines():
            if line:
                body = json.loads(line)
                print(body['response'], end='', flush=True)
                if 'done' in body and body['done']:
                    break
    except requests.exceptions.RequestException as e:
        print(f"Error during request: {e}")
    except FileNotFoundError:
        print(f"Error: Image file not found at {image_path}")
    except json.JSONDecodeError:
        print("Error: invalid json response")

# Example usage
image_path = "/mnt/nvme2n1/gemma/test.jpg"
prompt = "Describe what is in this image."
parse_image_with_gemma(image_path, prompt)

Relevant log output

[GIN] 2025/03/13 - 15:53:23 | 500 |  216.479399ms |       127.0.0.1 | POST     "/v1/chat/completions"
time=2025-03-13T15:53:24.713+08:00 level=INFO source=server.go:3634 msg="http: panic serving 127.0.0.1:34336: runtime error: integer divide by zero\ngoroutine 1321 [running]:\nnet/http.(*conn).serve.func1()\n\tnet/http/server.go:1947 +0xbe\npanic({0x5635f73af260?, 0x5635f7ca81c0?})\n\truntime/panic.go:787 +0x132\ngithub.com/ollama/ollama/model/models/gemma3.(*VisionModel).Forward(0xc0cf0e2870?, {0x5635f7506c00?, 0xc0cf0e2870?}, {0x5635f750f210?, 0xc0033aa630?})\n\tgithub.com/ollama/ollama/model/models/gemma3/model_vision.go:88 +0x399\ngithub.com/ollama/ollama/model/models/gemma3.(*Model).EncodeMultimodal(0xc000798070, {0x5635f7506c00, 0xc0cf0e2870}, {0xc12338c000, 0x227ec, 0x227ee})\n\tgithub.com/ollama/ollama/model/models/gemma3/model.go:106 +0x18e\ngithub.com/ollama/ollama/runner/ollamarunner.(*Server).inputs(0xc000034d80, {0x5635f7506c00, 0xc0cf0e2870}, {0xc000318320, 0xa0}, {0xc0cf0e2840, 0x1, 0xc0035d5650?})\n\tgithub.com/ollama/ollama/runner/ollamarunner/runner.go:191 +0x3d2\ngithub.com/ollama/ollama/runner/ollamarunner.(*Server).NewSequence(0xc000034d80, {0xc000318320, 0xa0}, {0xc0cf0e2840, 0x1, 0x1}, {0x8000, {0xc01ec6a3a0, 0x2, 0x2}, ...})\n\tgithub.com/ollama/ollama/runner/ollamarunner/runner.go:103 +0xd5\ngithub.com/ollama/ollama/runner/ollamarunner.(*Server).completion(0xc000034d80, {0x5635f74fd4f8, 0xc0cf0f6000}, 0xc003397b80)\n\tgithub.com/ollama/ollama/runner/ollamarunner/runner.go:602 +0x54f\nnet/http.HandlerFunc.ServeHTTP(0xc000762780?, {0x5635f74fd4f8?, 0xc0cf0f6000?}, 0xc0035d5b60?)\n\tnet/http/server.go:2294 +0x29\nnet/http.(*ServeMux).ServeHTTP(0x5635f61ca125?, {0x5635f74fd4f8, 0xc0cf0f6000}, 0xc003397b80)\n\tnet/http/server.go:2822 +0x1c4\nnet/http.serverHandler.ServeHTTP({0x5635f74f9b10?}, {0x5635f74fd4f8?, 0xc0cf0f6000?}, 0x1?)\n\tnet/http/server.go:3301 +0x8e\nnet/http.(*conn).serve(0xc001bfc3f0, {0x5635f74ff5a8, 0xc00078db90})\n\tnet/http/server.go:2102 +0x625\ncreated by net/http.(*Server).Serve in goroutine 1\n\tnet/http/server.go:3454 +0x485"

OS

Linux

GPU

Nvidia

CPU

Intel

Ollama version

0.6.0

Originally created by @Sisphyus on GitHub (Mar 13, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9717 ### What is the issue? (gemma) root@node19:/mnt/nvme2n1/gemma# ollama -v ollama version is 0.6.0 (gemma) root@node19:/mnt/nvme2n1/gemma# ollama list NAME ID SIZE MODIFIED hf.co/unsloth/gemma-3-27b-it-GGUF:Q4_K_M 3a14687fd09d 16 GB 2 hours ago test code ``` import requests import base64 import json def parse_image_with_gemma(image_path, prompt, ollama_url="http://localhost:8889/v1/chat/completions", model_name="hf.co/unsloth/gemma-3-27b-it-GGUF:Q4_K_M"): """ Sends an image and prompt to the Ollama API for processing. """ try: with open(image_path, "rb") as image_file: encoded_image = base64.b64encode(image_file.read()).decode("utf-8") payload = { "model": model_name, "messages": [ { "role": "user", "content": [ {"type": "text", "text": prompt}, { "type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{encoded_image}"}, }, ], } ], } response = requests.post(ollama_url, json=payload, stream=True) response.raise_for_status() # Raise an exception for bad status codes. for line in response.iter_lines(): if line: body = json.loads(line) print(body['response'], end='', flush=True) if 'done' in body and body['done']: break except requests.exceptions.RequestException as e: print(f"Error during request: {e}") except FileNotFoundError: print(f"Error: Image file not found at {image_path}") except json.JSONDecodeError: print("Error: invalid json response") # Example usage image_path = "/mnt/nvme2n1/gemma/test.jpg" prompt = "Describe what is in this image." parse_image_with_gemma(image_path, prompt) ``` ### Relevant log output ```shell [GIN] 2025/03/13 - 15:53:23 | 500 | 216.479399ms | 127.0.0.1 | POST "/v1/chat/completions" time=2025-03-13T15:53:24.713+08:00 level=INFO source=server.go:3634 msg="http: panic serving 127.0.0.1:34336: runtime error: integer divide by zero\ngoroutine 1321 [running]:\nnet/http.(*conn).serve.func1()\n\tnet/http/server.go:1947 +0xbe\npanic({0x5635f73af260?, 0x5635f7ca81c0?})\n\truntime/panic.go:787 +0x132\ngithub.com/ollama/ollama/model/models/gemma3.(*VisionModel).Forward(0xc0cf0e2870?, {0x5635f7506c00?, 0xc0cf0e2870?}, {0x5635f750f210?, 0xc0033aa630?})\n\tgithub.com/ollama/ollama/model/models/gemma3/model_vision.go:88 +0x399\ngithub.com/ollama/ollama/model/models/gemma3.(*Model).EncodeMultimodal(0xc000798070, {0x5635f7506c00, 0xc0cf0e2870}, {0xc12338c000, 0x227ec, 0x227ee})\n\tgithub.com/ollama/ollama/model/models/gemma3/model.go:106 +0x18e\ngithub.com/ollama/ollama/runner/ollamarunner.(*Server).inputs(0xc000034d80, {0x5635f7506c00, 0xc0cf0e2870}, {0xc000318320, 0xa0}, {0xc0cf0e2840, 0x1, 0xc0035d5650?})\n\tgithub.com/ollama/ollama/runner/ollamarunner/runner.go:191 +0x3d2\ngithub.com/ollama/ollama/runner/ollamarunner.(*Server).NewSequence(0xc000034d80, {0xc000318320, 0xa0}, {0xc0cf0e2840, 0x1, 0x1}, {0x8000, {0xc01ec6a3a0, 0x2, 0x2}, ...})\n\tgithub.com/ollama/ollama/runner/ollamarunner/runner.go:103 +0xd5\ngithub.com/ollama/ollama/runner/ollamarunner.(*Server).completion(0xc000034d80, {0x5635f74fd4f8, 0xc0cf0f6000}, 0xc003397b80)\n\tgithub.com/ollama/ollama/runner/ollamarunner/runner.go:602 +0x54f\nnet/http.HandlerFunc.ServeHTTP(0xc000762780?, {0x5635f74fd4f8?, 0xc0cf0f6000?}, 0xc0035d5b60?)\n\tnet/http/server.go:2294 +0x29\nnet/http.(*ServeMux).ServeHTTP(0x5635f61ca125?, {0x5635f74fd4f8, 0xc0cf0f6000}, 0xc003397b80)\n\tnet/http/server.go:2822 +0x1c4\nnet/http.serverHandler.ServeHTTP({0x5635f74f9b10?}, {0x5635f74fd4f8?, 0xc0cf0f6000?}, 0x1?)\n\tnet/http/server.go:3301 +0x8e\nnet/http.(*conn).serve(0xc001bfc3f0, {0x5635f74ff5a8, 0xc00078db90})\n\tnet/http/server.go:2102 +0x625\ncreated by net/http.(*Server).Serve in goroutine 1\n\tnet/http/server.go:3454 +0x485" ``` ### OS Linux ### GPU Nvidia ### CPU Intel ### Ollama version 0.6.0
GiteaMirror added the bug label 2026-04-12 17:51:51 -05:00
GiteaMirror commented 2026-04-12 17:51:54 -05:00
Author
Owner
Copy Link

@ItzCrazyKns commented on GitHub (Mar 13, 2025):

The GGUF files provided by Ollama have the image encoder model in them as well with default keys so Ollama can get the weights for image processing but the models on Hugging Face have a separate image encoder/projector so Ollama can't find those keys in the GGUF file and you see the error.

<!-- gh-comment-id:2721344206 --> @ItzCrazyKns commented on GitHub (Mar 13, 2025): The GGUF files provided by Ollama have the image encoder model in them as well with default keys so Ollama can get the weights for image processing but the models on Hugging Face have a separate image encoder/projector so Ollama can't find those keys in the GGUF file and you see the error.
GiteaMirror commented 2026-04-12 17:51:54 -05:00
Author
Owner
Copy Link

@dmatora commented on GitHub (Mar 14, 2025):

@ItzCrazyKns is there a way to get Q6 working in ollama? Q4 is too dumb, Q8 is too large to use full context in on 2x24Gb GPUs

<!-- gh-comment-id:2723747806 --> @dmatora commented on GitHub (Mar 14, 2025): @ItzCrazyKns is there a way to get Q6 working in ollama? Q4 is too dumb, Q8 is too large to use full context in on 2x24Gb GPUs
GiteaMirror commented 2026-04-12 17:51:54 -05:00
Author
Owner
Copy Link

@karaketir16 commented on GitHub (Mar 14, 2025):

The GGUF files provided by Ollama have the image encoder model in them as well with default keys so Ollama can get the weights for image processing but the models on Hugging Face have a separate image encoder/projector so Ollama can't find those keys in the GGUF file and you see the error.

@ItzCrazyKns

How can I download GGUF provided by Ollama so that I can run them on a machine without an internet connection?

<!-- gh-comment-id:2724941963 --> @karaketir16 commented on GitHub (Mar 14, 2025): > The GGUF files provided by Ollama have the image encoder model in them as well with default keys so Ollama can get the weights for image processing but the models on Hugging Face have a separate image encoder/projector so Ollama can't find those keys in the GGUF file and you see the error. @ItzCrazyKns How can I download GGUF provided by Ollama so that I can run them on a machine without an internet connection?
GiteaMirror commented 2026-04-12 17:51:55 -05:00
Author
Owner
Copy Link

@karaketir16 commented on GitHub (Mar 15, 2025):

How can I download GGUF provided by Ollama so that I can run them on a machine without an internet connection?

https://github.com/amirrezaDev1378/ollama-model-direct-download/

Okay, it is a great tool. But linux path seems incorrect for me, I have converted installing part to python.

https://gist.github.com/karaketir16/684a4d79465bc0a61eb613765c380f1e

<!-- gh-comment-id:2726569153 --> @karaketir16 commented on GitHub (Mar 15, 2025): > How can I download GGUF provided by Ollama so that I can run them on a machine without an internet connection? https://github.com/amirrezaDev1378/ollama-model-direct-download/ Okay, it is a great tool. But linux path seems incorrect for me, I have converted installing part to python. https://gist.github.com/karaketir16/684a4d79465bc0a61eb613765c380f1e
GiteaMirror commented 2026-04-12 17:51:55 -05:00
Author
Owner
Copy Link

@soldivelot commented on GitHub (Mar 22, 2025):

i have a abliterated llm from mlabonne/gemma-3-27b-it-abliterated-GGUF and a clip model from ggml-org/gemma-3-27b-it-GGUF, is there a tool to convert them to ollama's format?
is this format related to the new vision engine?

<!-- gh-comment-id:2745338384 --> @soldivelot commented on GitHub (Mar 22, 2025): i have a abliterated llm from mlabonne/gemma-3-27b-it-abliterated-GGUF and a clip model from ggml-org/gemma-3-27b-it-GGUF, is there a tool to convert them to ollama's format? is this format related to the new vision engine?
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#6350
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?