[GH-ISSUE #4714] In macOS Terminal.app, single Japanese character at the end of ongoing line disappears. #2972

Closed
opened 2026-04-12 13:21:15 -05:00 by GiteaMirror · 0 comments
Owner

Originally created by @tokyohandsome on GitHub (May 30, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/4714

Originally assigned to: @joshyan1 on GitHub.

What is the issue?

First, I'd like to thank you for fixing the Japanese and other multi-byte (double-width) character issues. Output of Japanese is much better than before.
However, while I'm testing a couple of Japanese LLM's I found another issue which does not seem to be related to LLM model.

When a sentence continues beyond the edge of the window, the last character gets removed. It's written then deleted.
Example:

>>> 東京にまつわる興味深い事実を教えてください。
東京にまつわる興味深い事実の一つに、「世界で最も人口密度が高い都市の一つであること」が挙げ 
れます。東京都特別区部は、約1,300万人もの人々が暮らしており、非常に狭い地域に多くの人々が生 
しています。

First line of response initially ended with "ら" but it's deleted when the line was wrapped and the following characters appeared in the next line.
Similarly, the second line initially had "活" at the end but it's gone when the sentence continued to the third line.
There are still a room for few more characters to be added at the end of each line.

It looks like Ollama does not cut or copy the full character code, i.e. only half byte of double-byte of a Japanese character. I assume so because there seems an invisible character or space ' ' exists instead of the missing character at the end of a line.

LLMs that I tested with (all resulted the same):
aya:35b-23-q4_0
andrewcanis/command-r:q4_0
ArrowPro-7B-KUJIRA-f16:converted (I converted from a gguf model)

OS

macOS

GPU

Apple

CPU

Apple

Ollama version

0.1.39

Originally created by @tokyohandsome on GitHub (May 30, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/4714 Originally assigned to: @joshyan1 on GitHub. ### What is the issue? First, I'd like to thank you for fixing the Japanese and other multi-byte (double-width) character issues. Output of Japanese is much better than before. However, while I'm testing a couple of Japanese LLM's I found another issue which does not seem to be related to LLM model. When a sentence continues beyond the edge of the window, the last character gets removed. It's written then deleted. Example: ``` >>> 東京にまつわる興味深い事実を教えてください。 東京にまつわる興味深い事実の一つに、「世界で最も人口密度が高い都市の一つであること」が挙げ れます。東京都特別区部は、約1,300万人もの人々が暮らしており、非常に狭い地域に多くの人々が生 しています。 ``` First line of response initially ended with "ら" but it's deleted when the line was wrapped and the following characters appeared in the next line. Similarly, the second line initially had "活" at the end but it's gone when the sentence continued to the third line. There are still a room for few more characters to be added at the end of each line. It looks like Ollama does not cut or copy the full character code, i.e. only half byte of double-byte of a Japanese character. I assume so because there seems an invisible character or space ' ' exists instead of the missing character at the end of a line. LLMs that I tested with (all resulted the same): aya:35b-23-q4_0 andrewcanis/command-r:q4_0 ArrowPro-7B-KUJIRA-f16:converted (I converted from a gguf model) ### OS macOS ### GPU Apple ### CPU Apple ### Ollama version 0.1.39
GiteaMirror added the bug label 2026-04-12 13:21:15 -05:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#2972