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 #14903] Better documentation needed for hf.co -> Modelfile workflow conversion and API call differences native/openai #56112

New Issue
Closed
opened 2026-04-29 10:16:37 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-29 10:16:37 -05:00
Owner
Copy Link

Originally created by @Xnyle on GitHub (Mar 17, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/14903

Trying to get different Qwen3.5 models from hf.co to run I'm currently stuck trying to disable thinking.

Available (at least the one I found) documentation / examples do not answer the following questions:

  1. Why does the standard Ollama Template for qwen3.5 just contain .Prompt and nothing else?
  2. If i use for instance unsloth/Qwen3.5-35B-A3B-GGUF it contains {%- if enable_thinking is defined and enable_thinking is false %} but how is that supposed to be set?
  3. what about the {{ if and .IsThinkSet (not .Think) -}} present in the qwen3 Templates (those can be set for instance via think:true in the OpenAi compatible completions API?
Originally created by @Xnyle on GitHub (Mar 17, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/14903 Trying to get different Qwen3.5 models from hf.co to run I'm currently stuck trying to disable thinking. Available (at least the one I found) documentation / examples do not answer the following questions: 1. Why does the standard Ollama Template for qwen3.5 just contain .Prompt and nothing else? 2. If i use for instance unsloth/Qwen3.5-35B-A3B-GGUF it contains {%- if enable_thinking is defined and enable_thinking is false %} but how is that supposed to be set? 3. what about the {{ if and $.IsThinkSet (not $.Think) -}} present in the qwen3 Templates (those can be set for instance via think:true in the OpenAi compatible completions API?
GiteaMirror added the feature request label 2026-04-29 10:16:37 -05:00
GiteaMirror commented 2026-04-29 10:16:39 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Mar 17, 2026):

  1. Template rendering and parsing for some models has been moved to dedicated code.
  2. That's the Jinja template which is not supported in ollama. It corresponds to the {{ if and $.IsThinkSet (not $.Think) -}} in the Go template used by ollama. This is controlled by the think parameter in the API call.
  3. Thinking control via the OpenAI is done by setting reasoning_effort in the API call.
<!-- gh-comment-id:4073818326 --> @rick-github commented on GitHub (Mar 17, 2026): 1. Template rendering and parsing for some models has been moved to [dedicated code](https://github.com/ollama/ollama/tree/main/model/parsers). 2. That's the Jinja template which is not supported in ollama. It corresponds to the `{{ if and $.IsThinkSet (not $.Think) -}}` in the Go template used by ollama. This is controlled by the [`think`](https://github.com/ollama/ollama/blob/main/docs/api.md#parameters-1:~:text=use%20if%20supported-,think,-%3A%20%28for%20thinking%20models) parameter in the API call. 3. Thinking control via the OpenAI is done by setting [`reasoning_effort`](https://github.com/ollama/ollama/blob/main/docs/api/openai-compatibility.mdx#supported-request-fields:~:text=%5Bx%5D-,reasoning_effort,-%28%22high%22) in the API call.
GiteaMirror commented 2026-04-29 10:16:41 -05:00
Author
Owner
Copy Link

@Xnyle commented on GitHub (Mar 17, 2026):

Thank you. Got a clearer picture now.

Suggestion so that not every newcomer trips over the same traps:

There should be an overview explaining how tokenizer.chat_template lands in the generated Modelfile, how that's then probably the wrong template format and so on.

Also for example https://docs.ollama.com/capabilities/thinking should mention that for /completions it's a different field/behaviour. More cross references between "Capabilities" and the two different APIs in general.

<!-- gh-comment-id:4074283379 --> @Xnyle commented on GitHub (Mar 17, 2026): Thank you. Got a clearer picture now. Suggestion so that not every newcomer trips over the same traps: There should be an overview explaining how tokenizer.chat_template lands in the generated Modelfile, how that's then probably the wrong template format and so on. Also for example https://docs.ollama.com/capabilities/thinking should mention that for /completions it's a different field/behaviour. More cross references between "Capabilities" and the two different APIs in general.
GiteaMirror commented 2026-04-29 10:16:42 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Mar 17, 2026):

https://github.com/ollama/ollama/blob/main/docs/modelfile.mdx#template

<!-- gh-comment-id:4074328763 --> @rick-github commented on GitHub (Mar 17, 2026): https://github.com/ollama/ollama/blob/main/docs/modelfile.mdx#template
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#56112
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?