github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-16 13:19:40 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #9652] Hard coding to not use cache #6298

New Issue
Closed
opened 2026-04-12 17:44:49 -05:00 by GiteaMirror · 4 comments
GiteaMirror commented 2026-04-12 17:44:49 -05:00
Owner
Copy Link

Originally created by @VistritPandey on GitHub (Mar 11, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9652

Originally assigned to: @jessegross on GitHub.

So in certain use cases, I do not have a specific example to share, it seems it relies on cached value. I am seeing this when using Llama-3.2-vision (11b). I am trying to process images, and for certain images, which are pretty similar looking, probably because of the encoding value, it seems to be using the cached value. Can I pass something in the JSON load that I am sending to Ollama, to not use cache, or can it be tied to the CLIP/embedding model? Thanks!

FWIW, it's happening on both Mac and Linux environments

Originally created by @VistritPandey on GitHub (Mar 11, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9652 Originally assigned to: @jessegross on GitHub. So in certain use cases, I do not have a specific example to share, it seems it relies on cached value. I am seeing this when using `Llama-3.2-vision (11b)`. I am trying to process images, and for certain images, which are pretty similar looking, probably because of the encoding value, it seems to be using the cached value. Can I pass something in the JSON load that I am sending to Ollama, to not use cache, or can it be tied to the CLIP/embedding model? Thanks! FWIW, it's happening on both Mac and Linux environments
GiteaMirror added the feature request label 2026-04-12 17:44:49 -05:00
GiteaMirror commented 2026-04-12 17:44:51 -05:00
Author
Owner
Copy Link

@jessegross commented on GitHub (Mar 17, 2025):

How are you testing that it looks like things are cached? Do you get the same results if you kill the Ollama server process vs. running /clear?

<!-- gh-comment-id:2730992106 --> @jessegross commented on GitHub (Mar 17, 2025): How are you testing that it looks like things are cached? Do you get the same results if you kill the Ollama server process vs. running `/clear`?
GiteaMirror commented 2026-04-12 17:44:53 -05:00
Author
Owner
Copy Link

@VistritPandey commented on GitHub (Mar 17, 2025):

I haven't tried running it with the /clear yet. But basically here are all the details:

  • I have the ollama docker running
  • The results I am testing are on a set of images, let's say 1000 in this case, which is basically like a batch being processed over some time like a day
  • There is no de-duplication, for the same reason that LLM might generate different responses, for the same set of images
  • I usually don't restart the server unless I am updating the docker | or even the app on mac

Let me know if you have any other questions! thank you!!

<!-- gh-comment-id:2731022505 --> @VistritPandey commented on GitHub (Mar 17, 2025): I haven't tried running it with the `/clear` yet. But basically here are all the details: - I have the ollama docker running - The results I am testing are on a set of images, let's say 1000 in this case, which is basically like a batch being processed over some time like a day - There is no de-duplication, for the same reason that LLM might generate different responses, for the same set of images - I usually don't restart the server unless I am updating the docker | or even the app on mac Let me know if you have any other questions! thank you!!
GiteaMirror commented 2026-04-12 17:44:55 -05:00
Author
Owner
Copy Link

@jessegross commented on GitHub (Mar 17, 2025):

How do you know that something is being cached? Even getting the same answer for similar images doesn't mean things aren't working right.

In order to find a root cause, we would need a more specific reproduction case and then try the things that I mentioned above.

<!-- gh-comment-id:2731125890 --> @jessegross commented on GitHub (Mar 17, 2025): How do you know that something is being cached? Even getting the same answer for similar images doesn't mean things aren't working right. In order to find a root cause, we would need a more specific reproduction case and then try the things that I mentioned above.
GiteaMirror commented 2026-04-12 17:44:58 -05:00
Author
Owner
Copy Link

@VistritPandey commented on GitHub (Mar 18, 2025):

Strangely, I haven't seen any cases to share since I restarted the docker yesterday. But feel free to close this issue, and I can reopen it if I encounter any instances. But thank you for looking into it!

<!-- gh-comment-id:2733442106 --> @VistritPandey commented on GitHub (Mar 18, 2025): Strangely, I haven't seen any cases to share since I restarted the docker yesterday. But feel free to close this issue, and I can reopen it if I encounter any instances. But thank you for looking into it!
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 feature request
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#6298
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?