github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-17 13:47:08 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #1567] getting llava output in chinese #62895

New Issue
Closed
opened 2026-05-03 10:40:19 -05:00 by GiteaMirror · 6 comments
GiteaMirror commented 2026-05-03 10:40:19 -05:00
Owner
Copy Link

Originally created by @YashKumarVerma on GitHub (Dec 16, 2023).
Original GitHub issue: https://github.com/ollama/ollama/issues/1567

I've tried running in CLI and via network request, but getting reply in chinese mandarin.

Has someone faced this?
image
image

Originally created by @YashKumarVerma on GitHub (Dec 16, 2023). Original GitHub issue: https://github.com/ollama/ollama/issues/1567 I've tried running in CLI and via network request, but getting reply in chinese mandarin. Has someone faced this? <img width="1376" alt="image" src="https://github.com/jmorganca/ollama/assets/14032427/bd80327a-e42e-45c2-9ad0-4222b6221bb7"> <img width="1123" alt="image" src="https://github.com/jmorganca/ollama/assets/14032427/af4b98bb-11f5-4ced-8d75-9f4ade11ee26">
GiteaMirror added the bug label 2026-05-03 10:40:19 -05:00
GiteaMirror commented 2026-05-03 10:40:23 -05:00
Author
Owner
Copy Link

@Yumcoder-dev commented on GitHub (Dec 17, 2023):

when I prompt "Describe the image in detail. /Users/..../figure-8-2.jpg" in the cmd everything works as expected.
But, when running the prompt from langchain (3.9 and 3.10, 3.11) the result was printed in China and was not related to the provided image.

the code :

from langchain.prompts import PromptTemplate

model_img = ChatOllama(model="llava:13b")
model_img.invoke("Describe the image in detail. /Users/..../figure-8-2.jpg")

AIMessage(content='在这张图片中，你可以看到一个室内场景，有一位女子正在拿着一个小男孩的手。他们似乎在一起参观一个博物馆或者是游览一个室内展览。\n\n场景中还有两位其他人，一人位于女子和小男孩的左侧，另一人则在他们的右侧。这些人都看起来像是游客或者是博物馆参观者。\n\n此外，还有一个手提包在场景中的下方，可能是女子或其他游客的物品。', additional_kwargs={}, example=False)

test with ollama 0.1.{14,15,16}

<!-- gh-comment-id:1859032213 --> @Yumcoder-dev commented on GitHub (Dec 17, 2023): when I prompt "Describe the image in detail. /Users/..../figure-8-2.jpg" in the cmd everything works as expected. But, when running the prompt from langchain (3.9 and 3.10, 3.11) the result was printed in China and was not related to the provided image. the code : ``` from langchain.prompts import PromptTemplate model_img = ChatOllama(model="llava:13b") model_img.invoke("Describe the image in detail. /Users/..../figure-8-2.jpg") ``` AIMessage(content='在这张图片中，你可以看到一个室内场景，有一位女子正在拿着一个小男孩的手。他们似乎在一起参观一个博物馆或者是游览一个室内展览。\n\n场景中还有两位其他人，一人位于女子和小男孩的左侧，另一人则在他们的右侧。这些人都看起来像是游客或者是博物馆参观者。\n\n此外，还有一个手提包在场景中的下方，可能是女子或其他游客的物品。', additional_kwargs={}, example=False) test with ollama 0.1.{14,15,16}
GiteaMirror commented 2026-05-03 10:40:24 -05:00
Author
Owner
Copy Link

@ndjstn commented on GitHub (Dec 19, 2023):

I am having a similar issue, not only do I gt Chinese, it is confused.

抱歉，我无法识别该文件的内容。请提供更多信息或上传其他图片以便我可以为您提供有关该图像的分析和解释。

total duration:       12.775150416s
load duration:        8.967403808s
prompt eval count:    31 token(s)
prompt eval duration: 264.066ms
prompt eval rate:     117.39 tokens/s
eval count:           70 token(s)
eval duration:        3.50308s
eval rate:            19.98 tokens/s

When using the model initially I tried:

ollama pull llava
ollama run llava --verbose "describe this pictur ~/Downloads/arm-studio-body.jpg"

I got results back in Chinese. Then I tried:

ollama pull llava:13b

And did the same thing. I got one response in English, two in Hebrew, one in Japanese, and the rest in Chinese across various images. None of them seemed to be able to access the file system. I tried removing the llava:latest and keeping the llava:13b as it was the only one to produce English, and tried again. I got similar results. I have tried jpg and png files.

NAME ID SIZE MODIFIED
llama2:latest fe938a131f40 3.8 GB 19 hours ago
llama2-uncensored:latest 44040b922233 3.8 GB 19 hours ago
llava:13b 99e3bfdf4655 8.0 GB 2 minutes ago
llava:latest e4c3eb471fd8 4.5 GB 18 hours ago
mistral:latest 1ab49bc0b6a8 4.1 GB 18 hours ago
mistral-repro:latest ceddb6201a88 4.1 GB 14 hours ago

//////767//////////////// OS: Pop!_OS 22.04 LTS x86_64
//////7676767676////////////// Kernel: 6.5.6-76060506-generic
/////76767//7676767////////////// Uptime: 1 day, 16 hours, 6 mins
/////767676///76767/////////////// Packages: 2645 (dpkg), 31 (flatpak)
///////767676///76767.///7676/////// Shell: bash 5.1.16
/////////767676//76767///767676//////// Resolution: 1920x1080, 1920x1080, 1920x1080
//////////76767676767////76767///////// DE: GNOME 42.5
///////////76767676//////7676////////// WM: Mutter
////////////,7676,///////767/////////// WM Theme: Pop
/////////////*7676///////76//////////// Theme: Pop-dark [GTK2/3]
///////////////7676//////////////////// Icons: Pop [GTK2/3]
///////////////7676///767//////////// Terminal: gnome-terminal
//////////////////////'//////////// CPU: AMD Ryzen 7 3700X (16) @ 3.600GHz
//////.7676767676767676767,////// GPU: NVIDIA GeForce RTX 3060 Lite Hash Rate
/////767676767676767676767///// GPU: NVIDIA GeForce RTX 3060
/////////////////////////// Memory: 8841MiB / 32012MiB

<!-- gh-comment-id:1863030912 --> @ndjstn commented on GitHub (Dec 19, 2023): I am having a similar issue, not only do I gt Chinese, it is confused. ``` 抱歉，我无法识别该文件的内容。请提供更多信息或上传其他图片以便我可以为您提供有关该图像的分析和解释。 total duration: 12.775150416s load duration: 8.967403808s prompt eval count: 31 token(s) prompt eval duration: 264.066ms prompt eval rate: 117.39 tokens/s eval count: 70 token(s) eval duration: 3.50308s eval rate: 19.98 tokens/s ``` When using the model initially I tried: ``` ollama pull llava ollama run llava --verbose "describe this pictur ~/Downloads/arm-studio-body.jpg" ``` I got results back in Chinese. Then I tried: ``` ollama pull llava:13b ``` And did the same thing. I got one response in English, two in Hebrew, one in Japanese, and the rest in Chinese across various images. None of them seemed to be able to access the file system. I tried removing the llava:latest and keeping the llava:13b as it was the only one to produce English, and tried again. I got similar results. I have tried jpg and png files. NAME ID SIZE MODIFIED llama2:latest fe938a131f40 3.8 GB 19 hours ago llama2-uncensored:latest 44040b922233 3.8 GB 19 hours ago llava:13b 99e3bfdf4655 8.0 GB 2 minutes ago llava:latest e4c3eb471fd8 4.5 GB 18 hours ago mistral:latest 1ab49bc0b6a8 4.1 GB 18 hours ago mistral-repro:latest ceddb6201a88 4.1 GB 14 hours ago //////*767//////////////// OS: Pop!_OS 22.04 LTS x86_64 //////7676767676*////////////// Kernel: 6.5.6-76060506-generic /////76767//7676767////////////// Uptime: 1 day, 16 hours, 6 mins /////767676///*76767/////////////// Packages: 2645 (dpkg), 31 (flatpak) ///////767676///76767.///7676*/////// Shell: bash 5.1.16 /////////767676//76767///767676//////// Resolution: 1920x1080, 1920x1080, 1920x1080 //////////76767676767////76767///////// DE: GNOME 42.5 ///////////76767676//////7676////////// WM: Mutter ////////////,7676,///////767/////////// WM Theme: Pop /////////////*7676///////76//////////// Theme: Pop-dark [GTK2/3] ///////////////7676//////////////////// Icons: Pop [GTK2/3] ///////////////7676///767//////////// Terminal: gnome-terminal //////////////////////'//////////// CPU: AMD Ryzen 7 3700X (16) @ 3.600GHz //////.7676767676767676767,////// GPU: NVIDIA GeForce RTX 3060 Lite Hash Rate /////767676767676767676767///// GPU: NVIDIA GeForce RTX 3060 /////////////////////////// Memory: 8841MiB / 32012MiB
GiteaMirror commented 2026-05-03 10:40:25 -05:00
Author
Owner
Copy Link

@BruceMacD commented on GitHub (Dec 19, 2023):

Hi all, this may not be the case for everyone who experiences this issue, but if you encounter this the multimodal model may not be supported by your version of Ollama. Try updating if you're not on latest and running the model again.

<!-- gh-comment-id:1863216973 --> @BruceMacD commented on GitHub (Dec 19, 2023): Hi all, this may not be the case for everyone who experiences this issue, but if you encounter this the multimodal model may not be supported by your version of Ollama. Try updating if you're not on latest and running the model again.
GiteaMirror commented 2026-05-03 10:40:26 -05:00
Author
Owner
Copy Link

@sam1am commented on GitHub (Jan 13, 2024):

I think the issue is that the image is not being passed. It works in the interactive chat, but not in python, or when running the command from command line.

<!-- gh-comment-id:1890295861 --> @sam1am commented on GitHub (Jan 13, 2024): I think the issue is that the image is not being passed. It works in the interactive chat, but not in python, or when running the command from command line.
GiteaMirror commented 2026-05-03 10:40:27 -05:00
Author
Owner
Copy Link

@Harrolee commented on GitHub (Jan 25, 2024):

Updating Ollama resolved this behavior for me.

<!-- gh-comment-id:1909242281 --> @Harrolee commented on GitHub (Jan 25, 2024): Updating Ollama resolved this behavior for me.
GiteaMirror commented 2026-05-03 10:40:28 -05:00
Author
Owner
Copy Link

@mchiang0610 commented on GitHub (Mar 11, 2024):

Hi all, this should be fixed in the recent builds of ollama (0.1.28+). If you have older models, we might have changed the prompt templates to help prompt it better too (please try pulling the models again via ollama pull command)

Sorry about this experience! Closing it for now, but please reach out if you are still seeing problems. Thanks!

<!-- gh-comment-id:1989188269 --> @mchiang0610 commented on GitHub (Mar 11, 2024): Hi all, this should be fixed in the recent builds of ollama (0.1.28+). If you have older models, we might have changed the prompt templates to help prompt it better too (please try pulling the models again via `ollama pull` command) Sorry about this experience! Closing it for now, but please reach out if you are still seeing problems. Thanks!
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#62895
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?