github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 08:02:14 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #6638] Llama 3.1 8b not generating answers since past few days #4179

New Issue
Closed
opened 2026-04-12 15:06:31 -05:00 by GiteaMirror · 9 comments
GiteaMirror commented 2026-04-12 15:06:31 -05:00
Owner
Copy Link

Originally created by @ToshiKBhat on GitHub (Sep 4, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/6638

Originally assigned to: @jmorganca on GitHub.

What is the issue?

The llama 3.1 8b model was generating answers in my RAG app until a few days back. Now it says i cannot help with that even when i use a simple system prompt - you are a helpful assistant , use the context provided to you to answer the user questions.
The 70b model seems to work fine, I also noticed the 8b model was updated recently.

OS

Linux

GPU

Nvidia

CPU

No response

Ollama version

No response

Originally created by @ToshiKBhat on GitHub (Sep 4, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/6638 Originally assigned to: @jmorganca on GitHub. ### What is the issue? The llama 3.1 8b model was generating answers in my RAG app until a few days back. Now it says i cannot help with that even when i use a simple system prompt - you are a helpful assistant , use the context provided to you to answer the user questions. The 70b model seems to work fine, I also noticed the 8b model was updated recently. ### OS Linux ### GPU Nvidia ### CPU _No response_ ### Ollama version _No response_
GiteaMirror added the bug label 2026-04-12 15:06:31 -05:00
GiteaMirror commented 2026-04-12 15:06:34 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Sep 4, 2024):

Server logs will help in debugging.

<!-- gh-comment-id:2329642881 --> @rick-github commented on GitHub (Sep 4, 2024): [Server logs](https://github.com/ollama/ollama/blob/main/docs/troubleshooting.md#how-to-troubleshoot-issues) will help in debugging.
GiteaMirror commented 2026-04-12 15:06:35 -05:00
Author
Owner
Copy Link

@ea167 commented on GitHub (Sep 5, 2024):

Same here. I'll make a separate issue with test case

<!-- gh-comment-id:2330450627 --> @ea167 commented on GitHub (Sep 5, 2024): Same here. I'll make a separate issue with test case
GiteaMirror commented 2026-04-12 15:06:36 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Sep 5, 2024):

Hi folks, what size of a prompt are you providing? Would it be possible to provide an example prompt that reproduces this? Thanks so much, and sorry this happened.

<!-- gh-comment-id:2332172814 --> @jmorganca commented on GitHub (Sep 5, 2024): Hi folks, what size of a prompt are you providing? Would it be possible to provide an example prompt that reproduces this? Thanks so much, and sorry this happened.
GiteaMirror commented 2026-04-12 15:06:37 -05:00
Author
Owner
Copy Link

@ToshiKBhat commented on GitHub (Sep 5, 2024):

Example prompt: you are a helpful assistant, you have to answer user’s questions using the context provided to you and nothing else.

I have removed the nothing else part and tried as well.
I thought maybe i am providing too much context and reduced it but it is just equivalent to 1-2 pages of a pdf, well under the 128k context limit.

<!-- gh-comment-id:2332439290 --> @ToshiKBhat commented on GitHub (Sep 5, 2024): Example prompt: you are a helpful assistant, you have to answer user’s questions using the context provided to you and nothing else. I have removed the nothing else part and tried as well. I thought maybe i am providing too much context and reduced it but it is just equivalent to 1-2 pages of a pdf, well under the 128k context limit.
GiteaMirror commented 2026-04-12 15:06:37 -05:00
Author
Owner
Copy Link

@ToshiKBhat commented on GitHub (Sep 5, 2024):

The response seems to be consistent - I can’t assist with that request. I suggest reaching out to

<!-- gh-comment-id:2332443519 --> @ToshiKBhat commented on GitHub (Sep 5, 2024): The response seems to be consistent - I can’t assist with that request. I suggest reaching out to <the support team mentioned in RAG documents>
GiteaMirror commented 2026-04-12 15:06:38 -05:00
Author
Owner
Copy Link

@ToshiKBhat commented on GitHub (Sep 5, 2024):

I set the num_ctx option to high based on the solution you provided on another issue and can see answer generation now.

was the default num_ctx changed recently? If so how can I find out what the recent change was ?

Thanks for providing the solution @jmorganca , really appreciate it.

<!-- gh-comment-id:2332467840 --> @ToshiKBhat commented on GitHub (Sep 5, 2024): I set the num_ctx option to high based on the solution you provided on another issue and can see answer generation now. was the default num_ctx changed recently? If so how can I find out what the recent change was ? Thanks for providing the solution @jmorganca , really appreciate it.
GiteaMirror commented 2026-04-12 15:06:39 -05:00
Author
Owner
Copy Link

@JonathanGuo01 commented on GitHub (Sep 6, 2024):

I had the same problem yesterday on my RAG system, which led me to think that I shouldn't have updated to the latest version of ollama.

<!-- gh-comment-id:2333074523 --> @JonathanGuo01 commented on GitHub (Sep 6, 2024): I had the same problem yesterday on my RAG system, which led me to think that I shouldn't have updated to the latest version of ollama.
GiteaMirror commented 2026-04-12 15:06:41 -05:00
Author
Owner
Copy Link

@antonkoenig commented on GitHub (Sep 10, 2024):

I see frequent crashes for ollama (pulled 9 days ago) in general. The webui sometimes reports server error 500. Most of the time it just appears stuck. I don't have a log of the error.

<!-- gh-comment-id:2339374377 --> @antonkoenig commented on GitHub (Sep 10, 2024): I see frequent crashes for ollama (pulled 9 days ago) in general. The webui sometimes reports server error 500. Most of the time it just appears stuck. I don't have a log of the error.
GiteaMirror commented 2026-04-12 15:06:42 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Nov 17, 2024):

I will close this as this I believe this is from having too small of a num_ctx (which is captured in other issues). Thanks again for the issue

<!-- gh-comment-id:2481078233 --> @jmorganca commented on GitHub (Nov 17, 2024): I will close this as this I believe this is from having too small of a `num_ctx` (which is captured in other issues). Thanks again for the issue
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#4179
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?