[GH-ISSUE #3534] When I ask a longer question, the output is arbitrary #48690

Open
opened 2026-04-28 09:06:07 -05:00 by GiteaMirror · 3 comments
Owner

Originally created by @0sengseng0 on GitHub (Apr 8, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/3534

What is the issue?

When I ask longer questions, it's about 15K or so. And I've set enough tokens: /set parameter num_ctx 4096. The model outputs a combination of numbers and symbols. Why?
image

What did you expect to see?

The model outputs normally

Steps to reproduce

ollama run qwen14:latest

Are there any recent changes that introduced the issue?

No response

OS

Linux

Architecture

x86

Platform

No response

Ollama version

0.1.27

GPU

Nvidia

GPU info

4090 * 2

CPU

No response

Other software

No response

Originally created by @0sengseng0 on GitHub (Apr 8, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/3534 ### What is the issue? When I ask longer questions, it's about 15K or so. And I've set enough tokens: /set parameter num_ctx 4096. The model outputs a combination of numbers and symbols. Why? ![image](https://github.com/ollama/ollama/assets/73268510/ef6853ad-465d-4879-9bf0-c1045d603a5d) ### What did you expect to see? The model outputs normally ### Steps to reproduce ollama run qwen14:latest ### Are there any recent changes that introduced the issue? _No response_ ### OS Linux ### Architecture x86 ### Platform _No response_ ### Ollama version 0.1.27 ### GPU Nvidia ### GPU info 4090 * 2 ### CPU _No response_ ### Other software _No response_
GiteaMirror added the needs more infobug labels 2026-04-28 09:06:07 -05:00
Author
Owner

@datalee commented on GitHub (Apr 11, 2024):

me too

<!-- gh-comment-id:2048866555 --> @datalee commented on GitHub (Apr 11, 2024): me too
Author
Owner

@0sengseng0 commented on GitHub (Apr 11, 2024):

me too

我发现当我在命令行配置【/set parameter num_ctx 4096】 或者在api调用中添加num_ctx 参数时,都会报这个错误,不确定是用法不对,还是这个参数有问题。 不用这个参数就行了。我从结果看上去,即便是用默认值,好像也是没有做截断的(不确定)。

<!-- gh-comment-id:2048871682 --> @0sengseng0 commented on GitHub (Apr 11, 2024): > me too 我发现当我在命令行配置【/set parameter num_ctx 4096】 或者在api调用中添加num_ctx 参数时,都会报这个错误,不确定是用法不对,还是这个参数有问题。 不用这个参数就行了。我从结果看上去,即便是用默认值,好像也是没有做截断的(不确定)。
Author
Owner

@rick-github commented on GitHub (Jan 30, 2026):

$ ollama run qwen14:latest
pulling manifest 
Error: pull model manifest: file does not exist
<!-- gh-comment-id:3821182080 --> @rick-github commented on GitHub (Jan 30, 2026): ```console $ ollama run qwen14:latest pulling manifest Error: pull model manifest: file does not exist ```
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#48690