github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 16:40:08 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #6212] show --modelfile (still) doesn't properly quote MESSAGE statements #3881

New Issue
Closed
opened 2026-04-12 14:43:17 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-12 14:43:17 -05:00
Owner
Copy Link

Originally created by @Maltz42 on GitHub (Aug 7, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/6212

What is the issue?

The patch added to v0.3.3 for issue #6103 didn't work, and actually made the situation harder to mitigate with find/replace. Request that patch be rolled back, the issue be re-opened, and quoting of MESSAGE strings be revisited and tested more thoroughly. Thanks!

(Or let me know if I should copy my posts from that issue here and just start anew.)

OS

Windows

GPU

Nvidia

CPU

Intel

Ollama version

0.3.4

Originally created by @Maltz42 on GitHub (Aug 7, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/6212 ### What is the issue? The patch added to v0.3.3 for issue #6103 didn't work, and actually made the situation harder to mitigate with find/replace. Request that patch be rolled back, the issue be re-opened, and quoting of MESSAGE strings be revisited and tested more thoroughly. Thanks! (Or let me know if I should copy my posts from that issue here and just start anew.) ### OS Windows ### GPU Nvidia ### CPU Intel ### Ollama version 0.3.4
GiteaMirror added the bug label 2026-04-12 14:43:17 -05:00
GiteaMirror commented 2026-04-12 14:43:19 -05:00
Author
Owner
Copy Link

@mxyng commented on GitHub (Aug 7, 2024):

From the linked issue:

The messages need to be surrounded by triple-quotes, not single quotes.

Multi-line strings can be surrounded with 1x or 3x double quotes depending on usage. Which type depends on the content:

  • 1x quote if content does not contain "
  • 3x quote otherwise

Here's your example from the linked issue:

  1. Create model tmp with multiline messages:

    FROM gemma2:2b
MESSAGE user """write a humorous limerick"""
MESSAGE assistant """There once was a fellow named Pete,
Whose dancing was quite incomplete.
He spun and he fell,
And let out a yell,
Now his dancing's no longer so neat!"""

  2. Check the manifest for tmp contains this messages layer:

    $ jq '.layers[] | select(.mediaType == "application/vnd.ollama.image.messages")' ~/.ollama/models/manifests/registry.ollama.ai/library/tmp/latest
{
  "mediaType": "application/vnd.ollama.image.messages",
  "digest": "sha256:561b820248ab51aad2a7903c8c28f102c1e7026a9cdaa6a852b307808e71617e",
  "size": 243
}

    which contains this list of messages

    $ jq . ~/.ollama/models/blobs/sha256-561b820248ab51aad2a7903c8c28f102c1e7026a9cdaa6a852b307808e71617e
[
  {
    "role": "user",
    "content": "write a humorous limerick"
  },
  {
    "role": "assistant",
    "content": "There once was a fellow named Pete,\nWhose dancing was quite incomplete.\nHe spun and he fell,\nAnd let out a yell,\nNow his dancing's no longer so neat!"
  }
]

  3. ollama show --modelfile tmp produces this output:

    # Modelfile generated by "ollama show"
# To build a new Modelfile based on this, replace FROM with:
# FROM tmp:latest

FROM /Users/<user>/.ollama/models/blobs/sha256-7462734796d67c40ecec2ca98eddf970e171dbb6b370e43fd633ee75b69abe1b
TEMPLATE """{{- range $i, $_ := .Messages }}
{{- $last := eq (len (slice $.Messages $i)) 1 }}
{{- if or (eq .Role "user") (eq .Role "system") }}<start_of_turn>user
{{ .Content }}<end_of_turn>
{{ if $last }}<start_of_turn>model
{{ end }}
{{- else if eq .Role "assistant" }}<start_of_turn>model
{{ .Content }}{{ if not $last }}<end_of_turn>
{{ end }}
{{- end }}
{{- end }}"""
PARAMETER stop <start_of_turn>
PARAMETER stop <end_of_turn>
LICENSE """
<snip>
"""
MESSAGE user write a humorous limerick
MESSAGE assistant "There once was a fellow named Pete,
Whose dancing was quite incomplete.
He spun and he fell,
And let out a yell,
Now his dancing's no longer so neat!"

  4. Create a model tmp2 from this show output and check the manifest for tmp2 shows it uses the exact same messages layer

    $ jq '.layers[] | select(.mediaType == "application/vnd.ollama.image.messages")' ~/.ollama/models/manifests/registry.ollama.ai/library/tmp2/latest
{
  "mediaType": "application/vnd.ollama.image.messages",
  "digest": "sha256:561b820248ab51aad2a7903c8c28f102c1e7026a9cdaa6a852b307808e71617e",
  "size": 243
}
<!-- gh-comment-id:2272447615 --> @mxyng commented on GitHub (Aug 7, 2024): From the linked issue: > The messages need to be surrounded by triple-quotes, not single quotes. Multi-line strings can be surrounded with 1x or 3x double quotes depending on usage. Which type depends on the content: * 1x quote if content does not contain `"` * 3x quote otherwise Here's your example from the linked issue: 1. Create model `tmp` with multiline messages: ``` FROM gemma2:2b MESSAGE user """write a humorous limerick""" MESSAGE assistant """There once was a fellow named Pete, Whose dancing was quite incomplete. He spun and he fell, And let out a yell, Now his dancing's no longer so neat!""" ``` 2. Check the manifest for `tmp` contains this messages layer: ``` $ jq '.layers[] | select(.mediaType == "application/vnd.ollama.image.messages")' ~/.ollama/models/manifests/registry.ollama.ai/library/tmp/latest { "mediaType": "application/vnd.ollama.image.messages", "digest": "sha256:561b820248ab51aad2a7903c8c28f102c1e7026a9cdaa6a852b307808e71617e", "size": 243 } ``` which contains this list of messages ``` $ jq . ~/.ollama/models/blobs/sha256-561b820248ab51aad2a7903c8c28f102c1e7026a9cdaa6a852b307808e71617e [ { "role": "user", "content": "write a humorous limerick" }, { "role": "assistant", "content": "There once was a fellow named Pete,\nWhose dancing was quite incomplete.\nHe spun and he fell,\nAnd let out a yell,\nNow his dancing's no longer so neat!" } ] ``` 3. `ollama show --modelfile tmp` produces this output: ``` # Modelfile generated by "ollama show" # To build a new Modelfile based on this, replace FROM with: # FROM tmp:latest FROM /Users/<user>/.ollama/models/blobs/sha256-7462734796d67c40ecec2ca98eddf970e171dbb6b370e43fd633ee75b69abe1b TEMPLATE """{{- range $i, $_ := .Messages }} {{- $last := eq (len (slice $.Messages $i)) 1 }} {{- if or (eq .Role "user") (eq .Role "system") }}<start_of_turn>user {{ .Content }}<end_of_turn> {{ if $last }}<start_of_turn>model {{ end }} {{- else if eq .Role "assistant" }}<start_of_turn>model {{ .Content }}{{ if not $last }}<end_of_turn> {{ end }} {{- end }} {{- end }}""" PARAMETER stop <start_of_turn> PARAMETER stop <end_of_turn> LICENSE """ <snip> """ MESSAGE user write a humorous limerick MESSAGE assistant "There once was a fellow named Pete, Whose dancing was quite incomplete. He spun and he fell, And let out a yell, Now his dancing's no longer so neat!" ``` 4. Create a model `tmp2` from this `show` output and check the manifest for `tmp2` shows it uses the exact same messages layer ``` $ jq '.layers[] | select(.mediaType == "application/vnd.ollama.image.messages")' ~/.ollama/models/manifests/registry.ollama.ai/library/tmp2/latest { "mediaType": "application/vnd.ollama.image.messages", "digest": "sha256:561b820248ab51aad2a7903c8c28f102c1e7026a9cdaa6a852b307808e71617e", "size": 243 } ```
GiteaMirror commented 2026-04-12 14:43:20 -05:00
Author
Owner
Copy Link

@Maltz42 commented on GitHub (Aug 7, 2024):

Yeah, after digging into this further, using fresh sessions, I think some of the things I was experiencing in the other issue may have been corruption from saving and re-importing modelfiles that screwed up quotes even further. I'm unable to reproduce any of that now in v0.3.3. I'll chalk this up to having made a mess of things on my end then for now. Thanks for the detailed walk through.

<!-- gh-comment-id:2272631547 --> @Maltz42 commented on GitHub (Aug 7, 2024): Yeah, after digging into this further, using fresh sessions, I think some of the things I was experiencing in the other issue may have been corruption from saving and re-importing modelfiles that screwed up quotes even further. I'm unable to reproduce any of that now in v0.3.3. I'll chalk this up to having made a mess of things on my end then for now. Thanks for the detailed walk through.
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#3881
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?