github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 16:11:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #2997] Can I force ollama to produce shorter responses? #27599

New Issue
Closed
opened 2026-04-22 05:03:59 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-22 05:03:59 -05:00
Owner
Copy Link

Originally created by @Anirudh257 on GitHub (Mar 8, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/2997

Hi,
I want to use the llama2 model available in Ollama to produce shorter outputs. I want to use max_new_tokens, max_length
parameters in https://huggingface.co/docs/transformers/en/main_classes/text_generation. Can I prompt the LLM to generate shorter sequences while keeping the meaning same?

There are some approaches given in https://www.reddit.com/r/LocalLLaMA/comments/14k7f5w/any_way_to_limit_the_output_to_a_specific_line/ but they don't work well.

Originally created by @Anirudh257 on GitHub (Mar 8, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/2997 Hi, I want to use the llama2 model available in Ollama to produce shorter outputs. I want to use ``max_new_tokens``, ``max_length`` parameters in https://huggingface.co/docs/transformers/en/main_classes/text_generation. Can I prompt the LLM to generate shorter sequences while keeping the meaning same? There are some approaches given in https://www.reddit.com/r/LocalLLaMA/comments/14k7f5w/any_way_to_limit_the_output_to_a_specific_line/ but they don't work well.
GiteaMirror commented 2026-04-22 05:04:01 -05:00
Author
Owner
Copy Link

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

One easy way to shorten answers would be to create a new model based on your llama2 model of choice and define your brevity instructions in the Makefile, under the SYSTEM section.

For example, I used prompt instructions suitable for a SYSTEM section:

ollama run llama2 "Please tell me a joke"

>>> Why don't scientists trust atoms? Because they make up everything! 😂

ollama run llama2 "Please tell me a joke. Don't use more than 5 words"

>>> Sure, here is a short joke:

>>> Kangaroo walks into a bar.
<!-- gh-comment-id:1988610109 --> @aosan commented on GitHub (Mar 11, 2024): One easy way to shorten answers would be to create a new model based on your llama2 model of choice and define your brevity instructions in the Makefile, under the SYSTEM section. For example, I used prompt instructions suitable for a SYSTEM section: ``` ollama run llama2 "Please tell me a joke" >>> Why don't scientists trust atoms? Because they make up everything! 😂 ollama run llama2 "Please tell me a joke. Don't use more than 5 words" >>> Sure, here is a short joke: >>> Kangaroo walks into a bar. ```
GiteaMirror commented 2026-04-22 05:04:01 -05:00
Author
Owner
Copy Link

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

Hi there, you can pass num_predict to the API

{
  "options": {
    "num_predict": 25
  }
}

However, this will truncate messages, so you should also make sure to tell the LLM to answer with short responses in the prompt.

<!-- gh-comment-id:1989547517 --> @jmorganca commented on GitHub (Mar 11, 2024): Hi there, you can pass `num_predict` to the API ``` { "options": { "num_predict": 25 } } ``` However, this will truncate messages, so you should also make sure to tell the LLM to answer with short responses in the prompt.
GiteaMirror commented 2026-04-22 05:04:02 -05:00
Author
Owner
Copy Link

@Anirudh257 commented on GitHub (Mar 13, 2024):

Thanks, @aosan and @jmorganca. This was pretty helpful!

<!-- gh-comment-id:1995081530 --> @Anirudh257 commented on GitHub (Mar 13, 2024): Thanks, @aosan and @jmorganca. This was pretty helpful!
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#27599
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?