github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-08 17:49:24 -05:00
Code Issues 14.6k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #2342] Quantize and Ollama Model #1354

New Issue
Closed
opened 2026-04-12 11:11:28 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-12 11:11:28 -05:00
Owner
Copy Link

Originally created by @stealthier-ai on GitHub (Feb 4, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/2342

I need to quantize a full version of an Ollama model that I layered in new weights for a specialized use case. Is there a way to do that within Ollama? It seems like I need to clone Llama.cpp and quantize through that. There are also other ways to quantize GGUF files and then recreate an Ollama model file. Am I missing anything or is there a specific method I should be using?

Originally created by @stealthier-ai on GitHub (Feb 4, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/2342 I need to quantize a full version of an Ollama model that I layered in new weights for a specialized use case. Is there a way to do that within Ollama? It seems like I need to clone Llama.cpp and quantize through that. There are also other ways to quantize GGUF files and then recreate an Ollama model file. Am I missing anything or is there a specific method I should be using?
GiteaMirror commented 2026-04-12 11:11:29 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Feb 4, 2024):

Hey @stealthier-ai . There are some instructions on how to do this here. I'm guessing you probably want to follow the steps for manually converting your model, but you don't actually need to clone a copy of llama.cpp if you have ollama already cloned, as there is a copy in the llm/llama.cpp directory. You can just run make quantize in that directory to build the binary.

That said, the process is less than ideal, and I've been working on creating a new way to convert/quantize models to make this a lot easier.

<!-- gh-comment-id:1925562605 --> @pdevine commented on GitHub (Feb 4, 2024): Hey @stealthier-ai . There are some instructions on how to do this [here](https://github.com/ollama/ollama/blob/main/docs/import.md). I'm guessing you probably want to follow the steps for [manually converting](https://github.com/ollama/ollama/blob/main/docs/import.md#manually-converting--quantizing-models) your model, but you don't actually need to clone a copy of llama.cpp if you have ollama already cloned, as there is a copy in the `llm/llama.cpp` directory. You can just run `make quantize` in that directory to build the binary. That said, the process is less than ideal, and I've been working on creating a new way to convert/quantize models to make this a lot easier.
GiteaMirror commented 2026-04-12 11:11:30 -05:00
Author
Owner
Copy Link

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

Going to go ahead and close this. Please feel free to reopen if you need to.

<!-- gh-comment-id:1989662284 --> @bmizerany commented on GitHub (Mar 11, 2024): Going to go ahead and close this. Please feel free to reopen if you need to.
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
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#1354
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?