[GH-ISSUE #6872] GGGGG in pesponce #66379

Closed
opened 2026-05-04 03:19:25 -05:00 by GiteaMirror · 2 comments
Owner

Originally created by @air55555 on GitHub (Sep 19, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/6872

Originally assigned to: @dhiltgen on GitHub.

What is the issue?

---- command log
C:\Program Files\Far Manager>ollama run llama3.1:70b >>> name
GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG

/?
Available Commands:
/set Set session variables
/show Show model information
/load Load a session or model
/save Save your current session
/clear Clear session context
/bye Exit
/?, /help Help for a command
/? shortcuts Help for keyboard shortcuts

Use """ to begin a multi-line message.

/show
Available Commands:
/show info Show details for this model
/show license Show model license
/show modelfile Show Modelfile for this model
/show parameters Show parameters for this model
/show system Show system message
/show template Show prompt template

/show info
Model
parameters 70.6B
quantization Q4_0
arch llama
context length 131072
embedding length 8192

Parameters
stop "<|start_header_id|>"
stop "<|end_header_id|>"
stop "<|eot_id|>"

License

    LLAMA 3.1 COMMUNITY LICENSE AGREEMENT
    Llama 3.1 Version Release Date: July 23, 2024

---- server log server.log

OS

Windows

GPU

Nvidia

CPU

AMD

Ollama version

0.3.10

Originally created by @air55555 on GitHub (Sep 19, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/6872 Originally assigned to: @dhiltgen on GitHub. ### What is the issue? ---- command log C:\Program Files\Far Manager>ollama run llama3.1:70b >>> name GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG >>> /? Available Commands: /set Set session variables /show Show model information /load <model> Load a session or model /save <model> Save your current session /clear Clear session context /bye Exit /?, /help Help for a command /? shortcuts Help for keyboard shortcuts Use """ to begin a multi-line message. >>> /show Available Commands: /show info Show details for this model /show license Show model license /show modelfile Show Modelfile for this model /show parameters Show parameters for this model /show system Show system message /show template Show prompt template >>> /show info Model parameters 70.6B quantization Q4_0 arch llama context length 131072 embedding length 8192 Parameters stop "<|start_header_id|>" stop "<|end_header_id|>" stop "<|eot_id|>" License LLAMA 3.1 COMMUNITY LICENSE AGREEMENT Llama 3.1 Version Release Date: July 23, 2024 ---- server log [server.log](https://github.com/user-attachments/files/17054228/server.log) ### OS Windows ### GPU Nvidia ### CPU AMD ### Ollama version 0.3.10
GiteaMirror added the memorynvidiabug labels 2026-05-04 03:19:30 -05:00
Author
Owner

@dhiltgen commented on GitHub (Sep 20, 2024):

This may be a scenario where we're slightly overshooting memory and windows is paging some of it and leading to gibberish. As a workaround, you can try to set OLLAMA_GPU_OVERHEAD on the server to reduce the amount of layers we load. If this works, let us know how much overhead you had to set aside and how many layers it was able to load to start producing correct results.

<!-- gh-comment-id:2364754273 --> @dhiltgen commented on GitHub (Sep 20, 2024): This may be a scenario where we're slightly overshooting memory and windows is paging some of it and leading to gibberish. As a workaround, you can try to set `OLLAMA_GPU_OVERHEAD` on the server to reduce the amount of layers we load. If this works, let us know how much overhead you had to set aside and how many layers it was able to load to start producing correct results.
Author
Owner

@air55555 commented on GitHub (Sep 25, 2024):

Just made fresh ollama install + model pull and now its ok . Thanks !

<!-- gh-comment-id:2373243742 --> @air55555 commented on GitHub (Sep 25, 2024): Just made fresh ollama install + model pull and now its ok . Thanks !
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#66379