github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-15 04:34:16 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #4807] 'Deepseek-V2' model output mix language #3033

New Issue
Closed
opened 2026-04-12 13:27:07 -05:00 by GiteaMirror · 5 comments
GiteaMirror commented 2026-04-12 13:27:07 -05:00
Owner
Copy Link

Originally created by @gigascake on GitHub (Jun 4, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/4807

What is the issue?

I used '16b-lite-chat-f16' (31GB) model.

This model inference speed more than before vesion model.

But, Sometimes, output response context in chinese and mix request language.

and then continues output mix language.

this situation. using RAG, document upload, and then Querying, occur just Problem.

image

Check,

Plz. the solution.

Os : Fedora 39
GPU : nvidia A4000 * 4
CPU : amd threadripper 7980x

Thanks. always.

OS

Linux

GPU

Nvidia

CPU

AMD

Ollama version

0.1.40

Originally created by @gigascake on GitHub (Jun 4, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/4807 ### What is the issue? I used '16b-lite-chat-f16' (31GB) model. This model inference speed more than before vesion model. But, Sometimes, output response context in chinese and mix request language. and then continues output mix language. this situation. using RAG, document upload, and then Querying, occur just Problem. ![image](https://github.com/ollama/ollama/assets/36724511/3f043b3c-7da0-4665-9cc5-a647fbc90f5d) Check, Plz. the solution. Os : Fedora 39 GPU : nvidia A4000 * 4 CPU : amd threadripper 7980x Thanks. always. ### OS Linux ### GPU Nvidia ### CPU AMD ### Ollama version 0.1.40
GiteaMirror added the bug label 2026-04-12 13:27:07 -05:00
GiteaMirror commented 2026-04-12 13:27:11 -05:00
Author
Owner
Copy Link

@gigascake commented on GitHub (Jun 4, 2024):

May Be, multi-language not support problem.

<!-- gh-comment-id:2146829114 --> @gigascake commented on GitHub (Jun 4, 2024): May Be, multi-language not support problem.
GiteaMirror commented 2026-04-12 13:27:12 -05:00
Author
Owner
Copy Link

@OldishCoder commented on GitHub (Jun 13, 2024):

I am seeing similar weirdness with deepseek-v2:16b. Sometimes it will only generate in Chinese. Even if I ask it to generate in English.

<!-- gh-comment-id:2165818104 --> @OldishCoder commented on GitHub (Jun 13, 2024): I am seeing similar weirdness with deepseek-v2:16b. Sometimes it will only generate in Chinese. Even if I ask it to generate in English.
GiteaMirror commented 2026-04-12 13:27:12 -05:00
Author
Owner
Copy Link

@itsPreto commented on GitHub (Jun 19, 2024):

I pulled deepseek-v2:16b yesterday and literally have not been able to get it to stop generating in Chinese... But I've not had any issues using the model on their free inferencing server: https://chat.deepseek.com/

could this be a gguf-conversion issue? curious to see if anyone with sufficient hardware has been able to test the f16 version

<!-- gh-comment-id:2179286155 --> @itsPreto commented on GitHub (Jun 19, 2024): I pulled [deepseek-v2:16b](https://ollama.com/library/deepseek-v2:16b) yesterday and literally have not been able to get it to stop generating in Chinese... But I've not had any issues using the model on their free inferencing server: https://chat.deepseek.com/ could this be a gguf-conversion issue? curious to see if anyone with sufficient hardware has been able to test the f16 version
GiteaMirror commented 2026-04-12 13:27:13 -05:00
Author
Owner
Copy Link

@gigascake commented on GitHub (Jun 20, 2024):

I have continuously tested it and there is no problem with the initial response. However, a problem arises in the second sentence. This is probably an issue where only Chinese is recognized as the memory context, and it becomes a problem if other languages ​​are in memory.

<!-- gh-comment-id:2179725503 --> @gigascake commented on GitHub (Jun 20, 2024): I have continuously tested it and there is no problem with the initial response. However, a problem arises in the second sentence. This is probably an issue where only Chinese is recognized as the memory context, and it becomes a problem if other languages ​​are in memory.
GiteaMirror commented 2026-04-12 13:27:13 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Jun 21, 2024):

Hi folks, we had a small correction to the template in https://ollama.com/library/deepseek-coder-v2 and https://ollama.com/library/deepseek-v2 - it should be fixed now and by default will answer in the prompt's language. Sorry about that!

<!-- gh-comment-id:2181782309 --> @jmorganca commented on GitHub (Jun 21, 2024): Hi folks, we had a small correction to the template in https://ollama.com/library/deepseek-coder-v2 and https://ollama.com/library/deepseek-v2 - it should be fixed now and by default will answer in the prompt's language. Sorry about that!
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#3033
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?