github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-16 05:05:37 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #858] Recreating the models pushed on ollama model registry #414

New Issue
Closed
opened 2026-04-12 10:04:30 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-12 10:04:30 -05:00
Owner
Copy Link

Originally created by @avinish1 on GitHub (Oct 20, 2023).
Original GitHub issue: https://github.com/ollama/ollama/issues/858

This issue is to understand about the process carried out for creating currently quantized models on the ollama registry like
https://ollama.ai/library/llama2 with any tag,

I have been trying to quantize a llama7b text generation model(https://huggingface.co/meta-llama/Llama-2-7b) from hugging face to have similiar performance to the one currently in ollama model registry.

Issue is the performance of quantized model from hugging face is not anywhere near the one which is in the ollama registry.

I have checked the following things:-

  1. Size of both the quantized models is same.
  2. I am using the same Modelfile to load the quantized model, this was by using the "Show" command for the llama2:latest model after pulling it.

Are there some more steps involved here to get the quantized model work same/similiar to the one currently already pushed to the registry(llama2:latest).

Originally created by @avinish1 on GitHub (Oct 20, 2023). Original GitHub issue: https://github.com/ollama/ollama/issues/858 This issue is to understand about the process carried out for creating currently quantized models on the ollama registry like https://ollama.ai/library/llama2 with any tag, I have been trying to quantize a llama7b text generation model(https://huggingface.co/meta-llama/Llama-2-7b) from hugging face to have similiar performance to the one currently in ollama model registry. Issue is the performance of quantized model from hugging face is not anywhere near the one which is in the ollama registry. I have checked the following things:- 1. Size of both the quantized models is same. 2. I am using the same Modelfile to load the quantized model, this was by using the "Show" command for the llama2:latest model after pulling it. Are there some more steps involved here to get the quantized model work same/similiar to the one currently already pushed to the registry(llama2:latest).
GiteaMirror added the question label 2026-04-12 10:04:30 -05:00
GiteaMirror commented 2026-04-12 10:04:32 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Oct 20, 2023):

Hi @avinish1, these steps are used: https://github.com/jmorganca/ollama/blob/main/docs/import.md. There isn't anything additional – what kind of performance difference are you seeing? Just to confirm, make sure you're using q4_0 (4 bit quantization) – that's what the default or latest tags use (e.g. ollama run llama2)

I'll close this for now but feel free to re-open if this doesn't help :)

<!-- gh-comment-id:1773291989 --> @jmorganca commented on GitHub (Oct 20, 2023): Hi @avinish1, these steps are used: https://github.com/jmorganca/ollama/blob/main/docs/import.md. There isn't anything additional – what kind of performance difference are you seeing? Just to confirm, make sure you're using `q4_0` (4 bit quantization) – that's what the default or `latest` tags use (e.g. `ollama run llama2`) I'll close this for now but feel free to re-open if this doesn't help :)
GiteaMirror commented 2026-04-12 10:04:32 -05:00
Author
Owner
Copy Link

@avinish1 commented on GitHub (Oct 20, 2023):

Thanks @jmorganca , have tried q4_0, let me come up with some examples on this and will reopen if needed.

<!-- gh-comment-id:1773323839 --> @avinish1 commented on GitHub (Oct 20, 2023): Thanks @jmorganca , have tried q4_0, let me come up with some examples on this and will reopen if needed.
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 question
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#414
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?