github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-05 23:53:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #5952] find system prompt encapsulation error in mistral-nemo 12b #3718

New Issue
Closed
opened 2026-04-12 14:31:48 -05:00 by GiteaMirror · 4 comments
GiteaMirror commented 2026-04-12 14:31:48 -05:00
Owner
Copy Link

Originally created by @map9 on GitHub (Jul 25, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/5952

What is the issue?

I used autogen + Ollama + mistral-nemo 12b model,
I find Ollama missed system message or lost question message.
maybe mistral-nemo 12b model template defined error.

case 1:

extractor_system_message = "...extractor_system_message..."
extractor = AssistantAgent(
"Extractor",
system_message = extractor_system_message,
llm_config = llm_config,
human_input_mode = "NEVER",
)

messages = [{"content": "...message...", "role": "user", "name": "Initializer"}]

ollama output:
time=2024-07-25T21:44:27.533+08:00 level=DEBUG source=routes.go:1337 msg="chat request" images=0 prompt="[INST]...message...[/INST]"

error is lost system prompt

case 2:

extractor_system_message = "...extractor_system_message..."
extractor = AssistantAgent(
"Extractor",
system_message = extractor_system_message,
llm_config = llm_config,
human_input_mode = "NEVER",
)

messages = [{"content": "...message1...", "role": "user", "name": "Initializer"},
{"content": "...message2...", "role": "user", "name": "Extractor"},
{"content": "...message3...", "role": "user", "name": "Editor"}]

ollama output:
time=2024-07-25T21:44:39.481+08:00 level=DEBUG source=routes.go:1337 msg="chat request" images=0 prompt="[INST] ...extractor_system_message...\n\n\n...message3...[/INST]"

error is lost message1, message2.

OS

macOS

GPU

Apple

CPU

Apple

Ollama version

0.2.8

Originally created by @map9 on GitHub (Jul 25, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/5952 ### What is the issue? I used autogen + Ollama + mistral-nemo 12b model, I find Ollama missed system message or lost question message. maybe mistral-nemo 12b model template defined error. case 1: ----------------------------------- extractor_system_message = "...extractor_system_message..." extractor = AssistantAgent( "Extractor", system_message = extractor_system_message, llm_config = llm_config, human_input_mode = "NEVER", ) messages = [{"content": "...message...", "role": "user", "name": "Initializer"}] ollama output: time=2024-07-25T21:44:27.533+08:00 level=DEBUG source=routes.go:1337 msg="chat request" images=0 prompt="[INST]...message...[/INST]" ------------------------------- error is lost system prompt case 2: ----------------------------------- extractor_system_message = "...extractor_system_message..." extractor = AssistantAgent( "Extractor", system_message = extractor_system_message, llm_config = llm_config, human_input_mode = "NEVER", ) messages = [{"content": "...message1...", "role": "user", "name": "Initializer"}, {"content": "...message2...", "role": "user", "name": "Extractor"}, {"content": "...message3...", "role": "user", "name": "Editor"}] ollama output: time=2024-07-25T21:44:39.481+08:00 level=DEBUG source=routes.go:1337 msg="chat request" images=0 prompt="[INST] ...extractor_system_message...\n\n\n...message3...[/INST]" ------------------------------- error is lost message1, message2. ### OS macOS ### GPU Apple ### CPU Apple ### Ollama version 0.2.8
GiteaMirror added the bug label 2026-04-12 14:31:48 -05:00
GiteaMirror commented 2026-04-12 14:31:49 -05:00
Author
Owner
Copy Link

@map9 commented on GitHub (Jul 25, 2024):

I find same error in llama3.1, glm4 models

<!-- gh-comment-id:2250425298 --> @map9 commented on GitHub (Jul 25, 2024): I find same error in llama3.1, glm4 models
GiteaMirror commented 2026-04-12 14:31:49 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Jul 25, 2024):

Ollama server logs may make it easier to diagnose the issue.

<!-- gh-comment-id:2250434981 --> @rick-github commented on GitHub (Jul 25, 2024): Ollama server logs may make it easier to diagnose the issue.
GiteaMirror commented 2026-04-12 14:31:50 -05:00
Author
Owner
Copy Link

@map9 commented on GitHub (Jul 25, 2024):

OK, I find input messages truncated by 2048 num_ctx.

<!-- gh-comment-id:2250594799 --> @map9 commented on GitHub (Jul 25, 2024): OK, I find input messages truncated by 2048 num_ctx.
GiteaMirror commented 2026-04-12 14:31:50 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Jul 25, 2024):

You can either adjust the API call to add "options": {"num_ctx": 8192}, or create a new model with a different default context size:

$ ollama show --modelfile mistral-nemo:12b-instruct-2407-q4_K_M > Modelfile
# edit Modefile and add "PARAMETER num_ctx 8192"
$ ollama create my-mistral-nemo -f Modelfile

Change 8192 to whatever size you need but be aware that the larger the context window, the more VRAM/RAM is needed.

<!-- gh-comment-id:2250664206 --> @rick-github commented on GitHub (Jul 25, 2024): You can either adjust the API call to add `"options": {"num_ctx": 8192}`, or create a new model with a different default context size: ```sh $ ollama show --modelfile mistral-nemo:12b-instruct-2407-q4_K_M > Modelfile # edit Modefile and add "PARAMETER num_ctx 8192" $ ollama create my-mistral-nemo -f Modelfile ``` Change `8192` to whatever size you need but be aware that the larger the context window, the more VRAM/RAM is needed.
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#3718
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?