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 #8099] ollama run silently truncating prompt #51689

New Issue
Closed
opened 2026-04-28 20:45:18 -05:00 by GiteaMirror · 4 comments
GiteaMirror commented 2026-04-28 20:45:18 -05:00
Owner
Copy Link

Originally created by @daniel-j-h on GitHub (Dec 14, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/8099

What is the issue?

The documentation shows us how to use ollama run to summarize a file, see

$ ollama run llama3.2 "Summarize this file: $(cat README.md)"

https://github.com/ollama/ollama?tab=readme-ov-file#pass-the-prompt-as-an-argument

What's not obvious here is that by default the prompt (and therefore the file passed in) gets truncated to 2048 tokens.

There's a warning in the server's logs but it's not clear to users of the ollama run command

time=2024-12-14T18:01:09.338Z level=WARN source=runner.go:129 msg="truncating input prompt" limit=2048 prompt=3858 keep=5 new=2048

This behavior is not obvious to users and easy to run into without even realizing.

It seems like it's not possible to change this behavior currently with ollama run. What's the best way forward here? Can we add a parameter to ollama run or should ollama run issue a warning for users?

OS

Linux

GPU

Other

CPU

Intel

Ollama version

0.5.1

Originally created by @daniel-j-h on GitHub (Dec 14, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/8099 ### What is the issue? The documentation shows us how to use `ollama run` to summarize a file, see $ ollama run llama3.2 "Summarize this file: $(cat README.md)" https://github.com/ollama/ollama?tab=readme-ov-file#pass-the-prompt-as-an-argument What's not obvious here is that by default the prompt (and therefore the file passed in) gets truncated to 2048 tokens. There's a warning in the server's logs but it's not clear to users of the ollama run command > time=2024-12-14T18:01:09.338Z level=WARN source=runner.go:129 msg="truncating input prompt" limit=2048 prompt=3858 keep=5 new=2048 This behavior is not obvious to users and easy to run into without even realizing. It seems like it's not possible to change this behavior currently with ollama run. What's the best way forward here? Can we add a parameter to ollama run or should ollama run issue a warning for users? ### OS Linux ### GPU Other ### CPU Intel ### Ollama version 0.5.1
GiteaMirror added the bug label 2026-04-28 20:45:18 -05:00
GiteaMirror commented 2026-04-28 20:45:20 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Dec 14, 2024):

https://github.com/ollama/ollama/blob/main/docs/faq.md#how-can-i-specify-the-context-window-size

The CLI is a simple frontend to the API and has limited functionality, different clients are available to make better use of the server. But you are right that the limitations aren't better explained.

A workaround for the problem you highlight is to increase the context window of the model.

$ ollama run llama3.2
>>> /set parameter num_ctx 32768
Set parameter 'num_ctx' to '32768'
>>> /save llama3.2-32k
Created new model 'llama3.2-32k'
>>> /bye
$ ollama run llama3.2-32k "Summarize this file: $(cat README.md)"
This file appears to be a documentation or README page for the Ollama project, which is a framework for building and running large language models.
...
<!-- gh-comment-id:2543316682 --> @rick-github commented on GitHub (Dec 14, 2024): https://github.com/ollama/ollama/blob/main/docs/faq.md#how-can-i-specify-the-context-window-size The CLI is a simple frontend to the API and has limited functionality, [different clients](https://github.com/ollama/ollama?tab=readme-ov-file#community-integrations) are available to make better use of the server. But you are right that the limitations aren't better explained. A workaround for the problem you highlight is to increase the context window of the model. ```console $ ollama run llama3.2 >>> /set parameter num_ctx 32768 Set parameter 'num_ctx' to '32768' >>> /save llama3.2-32k Created new model 'llama3.2-32k' >>> /bye $ ollama run llama3.2-32k "Summarize this file: $(cat README.md)" This file appears to be a documentation or README page for the Ollama project, which is a framework for building and running large language models. ... ```
GiteaMirror commented 2026-04-28 20:45:21 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Dec 14, 2024):

Related: #7043

<!-- gh-comment-id:2543321528 --> @rick-github commented on GitHub (Dec 14, 2024): Related: #7043
GiteaMirror commented 2026-04-28 20:45:21 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Dec 17, 2024):

@daniel-j-h thanks for the issue. I definitely agree it's not easy to see right now when the context length is exceeded. I'm going to close this in favor of #7043 though.

<!-- gh-comment-id:2549441663 --> @pdevine commented on GitHub (Dec 17, 2024): @daniel-j-h thanks for the issue. I definitely agree it's not easy to see right now when the context length is exceeded. I'm going to close this in favor of #7043 though.
GiteaMirror commented 2026-04-28 20:45:21 -05:00
Author
Owner
Copy Link

@gevzak commented on GitHub (Mar 13, 2025):

Was this implemented at some point? I am unable to find documentation for it.

<!-- gh-comment-id:2721990074 --> @gevzak commented on GitHub (Mar 13, 2025): Was this implemented at some point? I am unable to find documentation for it.
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#51689
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?