github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #15273] ollama show --modelfile not working for MLX Models #56282

New Issue
Open
opened 2026-04-29 10:33:48 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-29 10:33:48 -05:00
Owner
Copy Link

Originally created by @JoeLoginIsAlreadyTaken on GitHub (Apr 3, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/15273

Originally assigned to: @pdevine on GitHub.

What is the issue?

I backup models by using "ollama show --modelfile".
This works still well for the GGUF Models. There the output looks like this:

`l4b9@MacStudio model_export % ollama show --modelfile qwen3.5:122b
# Modelfile generated by "ollama show"
# To build a new Modelfile based on this, replace FROM with:
# FROM qwen3.5:122b

FROM /Users/l4b9/.ollama/models/blobs/sha256-93c83617a40560a61cda911ee327efdb5b5fbd39caa8b777a4ec565c0af1af3d
TEMPLATE {{ .Prompt }}
RENDERER qwen3.5
PARSER qwen3.5
PARAMETER top_k 20
PARAMETER top_p 0.95
PARAMETER temperature 1
LICENSE """ Apache License`

If I do this on one of the MLX-Models the FROM value is always missing.
The Output looks like this:

`l4b9@MacStudio model_export % ollama show --modelfile qwen3.5:35b-a3b-int8
# Modelfile generated by "ollama show"
# To build a new Modelfile based on this, replace FROM with:
# FROM qwen3.5:35b-a3b-int8

FROM
TEMPLATE {{ .Prompt }}
RENDERER qwen3.5
PARSER qwen3.5
PARAMETER top_p 0.95
PARAMETER min_p 0
PARAMETER presence_penalty 1.5
PARAMETER repeat_penalty 1
PARAMETER temperature 1
PARAMETER top_k 20
LICENSE """ Apache License`

Relevant log output

(not output in the logs)

OS

macOS

GPU

Apple

CPU

Apple

Ollama version

0.20

Originally created by @JoeLoginIsAlreadyTaken on GitHub (Apr 3, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/15273 Originally assigned to: @pdevine on GitHub. ### What is the issue? I backup models by using "ollama show --modelfile". This works still well for the GGUF Models. There the output looks like this: `l4b9@MacStudio model_export % ollama show --modelfile qwen3.5:122b \# Modelfile generated by "ollama show" \# To build a new Modelfile based on this, replace FROM with: \# FROM qwen3.5:122b FROM /Users/l4b9/.ollama/models/blobs/sha256-93c83617a40560a61cda911ee327efdb5b5fbd39caa8b777a4ec565c0af1af3d TEMPLATE {{ .Prompt }} RENDERER qwen3.5 PARSER qwen3.5 PARAMETER top_k 20 PARAMETER top_p 0.95 PARAMETER temperature 1 LICENSE """ Apache License` If I do this on one of the MLX-Models the FROM value is always missing. The Output looks like this: `l4b9@MacStudio model_export % ollama show --modelfile qwen3.5:35b-a3b-int8 \# Modelfile generated by "ollama show" \# To build a new Modelfile based on this, replace FROM with: \# FROM qwen3.5:35b-a3b-int8 FROM TEMPLATE {{ .Prompt }} RENDERER qwen3.5 PARSER qwen3.5 PARAMETER top_p 0.95 PARAMETER min_p 0 PARAMETER presence_penalty 1.5 PARAMETER repeat_penalty 1 PARAMETER temperature 1 PARAMETER top_k 20 LICENSE """ Apache License` ### Relevant log output ```shell (not output in the logs) ``` ### OS macOS ### GPU Apple ### CPU Apple ### Ollama version 0.20
GiteaMirror added the bug label 2026-04-29 10:33:48 -05:00
GiteaMirror commented 2026-04-29 10:33:49 -05:00
Author
Owner
Copy Link

@DrPiwo commented on GitHub (Apr 6, 2026):

+1

Even after fixing the modelfile, the creation of a new version of this model does not work.

% cat 1.modelfile

FROM qwen3.5:35b-a3b-nvfp4

PARAMETER min_p 0
PARAMETER presence_penalty 1.5
PARAMETER repeat_penalty 1
PARAMETER temperature 1
PARAMETER top_k 20
PARAMETER top_p 0.95

% ollama create NEW -f 1.modelfile

using existing layer sha256:dbe8c7ae5de2de6db8ee7d1539fc37c8541af265bd27eb0153c923cd80b42875
using existing layer sha256:a56d7308b8a6f7a7983deaf61f394b448cb072dd4d2d6097246a3691d3c0f5b6
using existing layer sha256:141bb5d142c76b3c9ec714b6ff8a05cac1e05a67c2e0a08a4f9170bcd202b08f
using existing layer sha256:e467a57bdb954f7d64984844f067f5e5df79557fb46d8a3ef42a0ae23332bf4f
using existing layer sha256:0291940ec5fc3828a519da2f57375bb095a1a0cceea1b53df056b8e56f615002
using existing layer sha256:c3e657208c66432b6b89a5070e13ca2fb56f385257dff45eb10fdcd545c3a10e
using existing layer sha256:21b4a418b6883b294ccd594d13a9dd0297daf3f0fdd1029185bf823d668c0e21
using existing layer sha256:a6cc00db7b716011f88450f57f57815e2a847572fdfc0f4c9118cb78b636e04a
using existing layer sha256:3c1cc3595658b48e23fa7c29cc00d55b374faaba5e77604e48daa40ab0d3cfae
using existing layer sha256:7324153458f3ded33f3d015307505efc6c850680f86fc281b87f050f863198d4
using existing layer sha256:00c0f31fc2e865dff6d112b150a7b4fe38d6fc1a61129d73c27686240ea5054e
using existing layer sha256:271b4dc157ade8256b18ea23e3b2e65b72fdb158acde5c63841965f5557aa458
using existing layer sha256:13eaac96b160a924ec62e25ff8ca07c134df8d9cbded806cce7da3233855b526
using existing layer sha256:5e4d7f74fec2f360eb9cfbfcd6ec0c4c76e684d3a11caaed259d9fd9bfbc7944
using existing layer sha256:4f25002776b741773666203dcea8f54619f177ace3ae483d311102092a4658e0
using existing layer sha256:27225450ac9c6529872ee1924fcb0962ff5634834f817040f444118116f4e516
using existing layer sha256:5f9e4d4901a92b997e463c1f46055088b6cca5ca61a6522d1b9f64c4bb81cb42
using existing layer sha256:316230d6a809701f4db5ea8f8fc862bc3a6f3229c937c174e674ff3ca0a64ac8
using existing layer sha256:7768af27c1fafa9cc9011c1dc20067e03f8915e03b63504550e11d5066986d13
using existing layer sha256:ce99b4cb2983d118806ce0a8b777a35b093e2000a503ebde25853284c9dfa003
using existing layer sha256:5f3a3c817e78f5b8a4ad2d2c458a3e4b2cce470d6c12642c4eeb12cb8a9bf51d
using existing layer sha256:86eff881e8d2c2b1caa3966ba775d007f76a693d8541d0973b1bd2aab4214a43
writing manifest
success

% ollama run NEW

pulling manifest
Error: pull model manifest: file does not exist

<!-- gh-comment-id:4191141173 --> @DrPiwo commented on GitHub (Apr 6, 2026): +1 Even after fixing the modelfile, the creation of a new version of this model does not work. % cat 1.modelfile ---- FROM qwen3.5:35b-a3b-nvfp4 PARAMETER min_p 0 PARAMETER presence_penalty 1.5 PARAMETER repeat_penalty 1 PARAMETER temperature 1 PARAMETER top_k 20 PARAMETER top_p 0.95 % ollama create NEW -f 1.modelfile ------- using existing layer sha256:dbe8c7ae5de2de6db8ee7d1539fc37c8541af265bd27eb0153c923cd80b42875 using existing layer sha256:a56d7308b8a6f7a7983deaf61f394b448cb072dd4d2d6097246a3691d3c0f5b6 using existing layer sha256:141bb5d142c76b3c9ec714b6ff8a05cac1e05a67c2e0a08a4f9170bcd202b08f using existing layer sha256:e467a57bdb954f7d64984844f067f5e5df79557fb46d8a3ef42a0ae23332bf4f using existing layer sha256:0291940ec5fc3828a519da2f57375bb095a1a0cceea1b53df056b8e56f615002 using existing layer sha256:c3e657208c66432b6b89a5070e13ca2fb56f385257dff45eb10fdcd545c3a10e using existing layer sha256:21b4a418b6883b294ccd594d13a9dd0297daf3f0fdd1029185bf823d668c0e21 using existing layer sha256:a6cc00db7b716011f88450f57f57815e2a847572fdfc0f4c9118cb78b636e04a using existing layer sha256:3c1cc3595658b48e23fa7c29cc00d55b374faaba5e77604e48daa40ab0d3cfae using existing layer sha256:7324153458f3ded33f3d015307505efc6c850680f86fc281b87f050f863198d4 using existing layer sha256:00c0f31fc2e865dff6d112b150a7b4fe38d6fc1a61129d73c27686240ea5054e using existing layer sha256:271b4dc157ade8256b18ea23e3b2e65b72fdb158acde5c63841965f5557aa458 using existing layer sha256:13eaac96b160a924ec62e25ff8ca07c134df8d9cbded806cce7da3233855b526 using existing layer sha256:5e4d7f74fec2f360eb9cfbfcd6ec0c4c76e684d3a11caaed259d9fd9bfbc7944 using existing layer sha256:4f25002776b741773666203dcea8f54619f177ace3ae483d311102092a4658e0 using existing layer sha256:27225450ac9c6529872ee1924fcb0962ff5634834f817040f444118116f4e516 using existing layer sha256:5f9e4d4901a92b997e463c1f46055088b6cca5ca61a6522d1b9f64c4bb81cb42 using existing layer sha256:316230d6a809701f4db5ea8f8fc862bc3a6f3229c937c174e674ff3ca0a64ac8 using existing layer sha256:7768af27c1fafa9cc9011c1dc20067e03f8915e03b63504550e11d5066986d13 using existing layer sha256:ce99b4cb2983d118806ce0a8b777a35b093e2000a503ebde25853284c9dfa003 using existing layer sha256:5f3a3c817e78f5b8a4ad2d2c458a3e4b2cce470d6c12642c4eeb12cb8a9bf51d using existing layer sha256:86eff881e8d2c2b1caa3966ba775d007f76a693d8541d0973b1bd2aab4214a43 writing manifest success % ollama run NEW ------ pulling manifest Error: pull model manifest: file does not exist
GiteaMirror commented 2026-04-29 10:33:50 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Apr 13, 2026):

@JoeLoginIsAlreadyTaken unfortunately backing up this way isn't going to work great with the MLX safetensors based models. The older GGUF based models use a single GGUF file which is references in the FROM line, but the new MLX based models reference a lot of different tensors which are each stored in a separate blob.

@DrPiwo Your issue is slightly different. If you want to create this again use ollama create --experimental instead of ollama create.

<!-- gh-comment-id:4234018524 --> @pdevine commented on GitHub (Apr 13, 2026): @JoeLoginIsAlreadyTaken unfortunately backing up this way isn't going to work great with the MLX safetensors based models. The older GGUF based models use a single GGUF file which is references in the FROM line, but the new MLX based models reference a lot of different tensors which are each stored in a separate blob. @DrPiwo Your issue is slightly different. If you want to create this again use `ollama create --experimental` instead of `ollama create`.
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#56282
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?