github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 16:40:08 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #5418] DeepSeek-Coder-V2 (Lite) spouts GGGs #3391

New Issue
Closed
opened 2026-04-12 14:01:31 -05:00 by GiteaMirror · 9 comments
GiteaMirror commented 2026-04-12 14:01:31 -05:00
Owner
Copy Link

Originally created by @lorenzodimauro97 on GitHub (Jul 1, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/5418

What is the issue?

Eventually when using deepseek-coder-v2:16b-lite-instruct-q8_0 and Open Web UI (but also other means like for example continuedev) the model will stop working and spout GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG regardless of the input:

image

The only fix is to force stop and restart the model, which is bothersome enough to be worth issuing a bug

OS

Windows

GPU

AMD

CPU

AMD

Ollama version

0.1.48

Originally created by @lorenzodimauro97 on GitHub (Jul 1, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/5418 ### What is the issue? Eventually when using deepseek-coder-v2:16b-lite-instruct-q8_0 and Open Web UI (but also other means like for example continuedev) the model will stop working and spout GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG regardless of the input: ![image](https://github.com/ollama/ollama/assets/50343905/b90c39ae-9269-4f7f-b3a5-dda65186a208) The only fix is to force stop and restart the model, which is bothersome enough to be worth issuing a bug ### OS Windows ### GPU AMD ### CPU AMD ### Ollama version 0.1.48
GiteaMirror added the bug label 2026-04-12 14:01:31 -05:00
GiteaMirror commented 2026-04-12 14:01:32 -05:00
Author
Owner
Copy Link

@igorschlum commented on GitHub (Jul 1, 2024):

Hi @lorenzodimauro97, with how many queries do the GGG appear?

<!-- gh-comment-id:2201233682 --> @igorschlum commented on GitHub (Jul 1, 2024): Hi @lorenzodimauro97, with how many queries do the GGG appear?
GiteaMirror commented 2026-04-12 14:01:33 -05:00
Author
Owner
Copy Link

@lorenzodimauro97 commented on GitHub (Jul 1, 2024):

It's not consistent, they can appear as soon as 2 queries or as long as about 30+, furthermore, there seems to be no errors on the console log

<!-- gh-comment-id:2201274364 --> @lorenzodimauro97 commented on GitHub (Jul 1, 2024): It's not consistent, they can appear as soon as 2 queries or as long as about 30+, furthermore, there seems to be no errors on the console log
GiteaMirror commented 2026-04-12 14:01:34 -05:00
Author
Owner
Copy Link

@igorschlum commented on GitHub (Jul 1, 2024):

BTW this model is the only one exept ChatGPT that can answer correctly this prompt:
how many days between febuary 2nd 2021 and may 23rd, 2024?
I will try to ask him more questions. I'm on a macbook with 32GB or RAM.

<!-- gh-comment-id:2201294854 --> @igorschlum commented on GitHub (Jul 1, 2024): BTW this model is the only one exept ChatGPT that can answer correctly this prompt: how many days between febuary 2nd 2021 and may 23rd, 2024? I will try to ask him more questions. I'm on a macbook with 32GB or RAM.
GiteaMirror commented 2026-04-12 14:01:34 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Jul 1, 2024):

Hi there, would it be possible to re-pull the model? This should fix things. Let me know if it continues though

<!-- gh-comment-id:2201329772 --> @jmorganca commented on GitHub (Jul 1, 2024): Hi there, would it be possible to re-pull the model? This should fix things. Let me know if it continues though
GiteaMirror commented 2026-04-12 14:01:35 -05:00
Author
Owner
Copy Link

@igorschlum commented on GitHub (Jul 1, 2024):

@jmorganca I played with this model since then and had no problem on MacOS.

<!-- gh-comment-id:2201446381 --> @igorschlum commented on GitHub (Jul 1, 2024): @jmorganca I played with this model since then and had no problem on MacOS.
GiteaMirror commented 2026-04-12 14:01:35 -05:00
Author
Owner
Copy Link

@lorenzodimauro97 commented on GitHub (Jul 2, 2024):

Hi there, would it be possible to re-pull the model? This should fix things. Let me know if it continues though

I did re-pull the model but problem still occurs:
image

<!-- gh-comment-id:2202664753 --> @lorenzodimauro97 commented on GitHub (Jul 2, 2024): > Hi there, would it be possible to re-pull the model? This should fix things. Let me know if it continues though I did re-pull the model but problem still occurs: ![image](https://github.com/ollama/ollama/assets/50343905/971dd3dd-15af-48e6-a7b7-d1826960fbf4)
GiteaMirror commented 2026-04-12 14:01:36 -05:00
Author
Owner
Copy Link

@igorschlum commented on GitHub (Jul 2, 2024):

I played quite a lot with this Model on MacOS without any problem. How much memory do you have? Can you write a simple Python script that can reproduce the issue so other users of Ollama could try to help on this issue?

<!-- gh-comment-id:2203125288 --> @igorschlum commented on GitHub (Jul 2, 2024): I played quite a lot with this Model on MacOS without any problem. How much memory do you have? Can you write a simple Python script that can reproduce the issue so other users of Ollama could try to help on this issue?
GiteaMirror commented 2026-04-12 14:01:36 -05:00
Author
Owner
Copy Link

@lorenzodimauro97 commented on GitHub (Jul 2, 2024):

I played quite a lot with this Model on MacOS without any problem. How much memory do you have? Can you write a simple Python script that can reproduce the issue so other users of Ollama could try to help on this issue?

I have 48 GB DDR4 at 3800 MHz, a Ryzen 5800X and an AMD 6800XT. I am using Windows 11.

I do not incur in memory saturation with the model, I stay at about 50-60% RAM usage, going 80-90 with heavy coding, but still, should be enough.

I do not have the means to reproduce it because it does not happen consistently. It can happen as often as after 3 queries or even after half a day's worth of coding, but still I find it hard to reproduce.

What I can say though is that it's the only model that has this issue, none of my other models seem to fail as hard and as frequently as this one does.

<!-- gh-comment-id:2203725611 --> @lorenzodimauro97 commented on GitHub (Jul 2, 2024): > I played quite a lot with this Model on MacOS without any problem. How much memory do you have? Can you write a simple Python script that can reproduce the issue so other users of Ollama could try to help on this issue? I have 48 GB DDR4 at 3800 MHz, a Ryzen 5800X and an AMD 6800XT. I am using Windows 11. I do not incur in memory saturation with the model, I stay at about 50-60% RAM usage, going 80-90 with heavy coding, but still, should be enough. I do not have the means to reproduce it because it does not happen consistently. It can happen as often as after 3 queries or even after half a day's worth of coding, but still I find it hard to reproduce. What I can say though is that it's the only model that has this issue, none of my other models seem to fail as hard and as frequently as this one does.
GiteaMirror commented 2026-04-12 14:01:36 -05:00
Author
Owner
Copy Link

@igorschlum commented on GitHub (Jul 2, 2024):

Thank you. Your graphics card has 16 GB of VRAM, but the model requires 17 GB of RAM, so it cannot fully fit in VRAM. Could you try testing a smaller model, such as ollama run deepseek-coder-v2:latest, and see if the issue with GGGGG still occurs? Then, test the deepseek-coder-v2:16b-lite-instruct-q5_0 model, which is 11 GB, to see if the problem can be reproduced.

<!-- gh-comment-id:2203994621 --> @igorschlum commented on GitHub (Jul 2, 2024): Thank you. Your graphics card has 16 GB of VRAM, but the model requires 17 GB of RAM, so it cannot fully fit in VRAM. Could you try testing a smaller model, such as ollama run deepseek-coder-v2:latest, and see if the issue with GGGGG still occurs? Then, test the deepseek-coder-v2:16b-lite-instruct-q5_0 model, which is 11 GB, to see if the problem can be reproduced.
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#3391
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?