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 #4862] Probably I am missing something... #3076

New Issue
Closed
opened 2026-04-12 13:30:57 -05:00 by GiteaMirror · 5 comments
GiteaMirror commented 2026-04-12 13:30:57 -05:00
Owner
Copy Link

Originally created by @Zibri on GitHub (Jun 6, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/4862

What is the issue?

I created a file containing: FROM I:\models\Mistral-7b-Instruct-v0.3.f16.q6_k.gguf

Then I did: ollama create mistral file
The model loaded.
then I did:
ollama run mistral

and If I say "Hello" it starts talking by itself introdfucing itself every time with a different identity.

Also, my GPU is an NVIDIA GTX 970M (which works with llama.cpp and opencl), but it's not apparently used.

OS

Windows

GPU

Nvidia

CPU

Intel

Ollama version

0.1.41

Originally created by @Zibri on GitHub (Jun 6, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/4862 ### What is the issue? I created a file containing: ``FROM I:\models\Mistral-7b-Instruct-v0.3.f16.q6_k.gguf`` Then I did: `ollama create mistral file` The model loaded. then I did: ollama run mistral and If I say "Hello" it starts talking by itself introdfucing itself every time with a different identity. Also, my GPU is an NVIDIA GTX 970M (which works with llama.cpp and opencl), but it's not apparently used. ### OS Windows ### GPU Nvidia ### CPU Intel ### Ollama version 0.1.41
GiteaMirror added the bug label 2026-04-12 13:30:57 -05:00
GiteaMirror commented 2026-04-12 13:30:58 -05:00
Author
Owner
Copy Link

@Zibri commented on GitHub (Jun 6, 2024):

Another example:

>>> ciao


"We're here because you wanted to learn how to paint, and I offered to help you," replied the woman with a
sigh. "You can't just give up after one try. Painting takes time and patience."

The young man looked at her with frustration in his eyes. "I've been trying for weeks now, and I still can't
make anything that looks decent. I don't have the talent for it."

"Talent is overrated," the woman said. "What you need is practice. Every great artist started out as a
beginner. You just have to keep trying and not give up."

The young man sighed again and looked back down at his canvas, still covered in splotches of unblended
paint. "I guess you're right," he said quietly. "But it's so hard."

"I know it is," the woman replied gently. "But remember why you wanted to learn how to paint in the first
place. What was it that drew you to it? That passion will carry you through the tough times."

The young man thought for a moment, then nodded slowly. "You're right," he said. "I just need to keep
going."

"That's the spirit!" the woman exclaimed, clapping her hands together. "Now let's get back to work and make
something amazing!"

The young man picked up his paintbrush and began to work on his canvas again, determined to improve. As they
worked side by side, the woman offered advice and encouragement, helping the young man to create a beautiful
painting that he could be proud of. It may have been difficult, but in the end, it was all worth it.

total duration:       1m42.4288454s
load duration:        1.6325ms
prompt eval count:    2 token(s)
prompt eval duration: 435.183ms
prompt eval rate:     4.60 tokens/s
eval count:           370 token(s)
eval duration:        1m41.988013s
eval rate:            3.63 tokens/s
<!-- gh-comment-id:2153011802 --> @Zibri commented on GitHub (Jun 6, 2024): Another example: ``` >>> ciao "We're here because you wanted to learn how to paint, and I offered to help you," replied the woman with a sigh. "You can't just give up after one try. Painting takes time and patience." The young man looked at her with frustration in his eyes. "I've been trying for weeks now, and I still can't make anything that looks decent. I don't have the talent for it." "Talent is overrated," the woman said. "What you need is practice. Every great artist started out as a beginner. You just have to keep trying and not give up." The young man sighed again and looked back down at his canvas, still covered in splotches of unblended paint. "I guess you're right," he said quietly. "But it's so hard." "I know it is," the woman replied gently. "But remember why you wanted to learn how to paint in the first place. What was it that drew you to it? That passion will carry you through the tough times." The young man thought for a moment, then nodded slowly. "You're right," he said. "I just need to keep going." "That's the spirit!" the woman exclaimed, clapping her hands together. "Now let's get back to work and make something amazing!" The young man picked up his paintbrush and began to work on his canvas again, determined to improve. As they worked side by side, the woman offered advice and encouragement, helping the young man to create a beautiful painting that he could be proud of. It may have been difficult, but in the end, it was all worth it. total duration: 1m42.4288454s load duration: 1.6325ms prompt eval count: 2 token(s) prompt eval duration: 435.183ms prompt eval rate: 4.60 tokens/s eval count: 370 token(s) eval duration: 1m41.988013s eval rate: 3.63 tokens/s ```
GiteaMirror commented 2026-04-12 13:30:58 -05:00
Author
Owner
Copy Link

@Zibri commented on GitHub (Jun 6, 2024):

How can I do a normal inference as I do with llama.cpp ?

<!-- gh-comment-id:2153012629 --> @Zibri commented on GitHub (Jun 6, 2024): How can I do a normal inference as I do with llama.cpp ?
GiteaMirror commented 2026-04-12 13:30:58 -05:00
Author
Owner
Copy Link

@Zibri commented on GitHub (Jun 6, 2024):

Hmm I think I found the reason. Creating a model as I did the template is not taken from the model but must be specified.
So I added /set template [INST] {{ if .System }}{{ .System }} {{ end }}{{ .Prompt }} [/INST]
And the system prompt worked.

Now the problem I have is that it talks too much :D
I wonder what paremeter limits the length of the responses without truncating them...

<!-- gh-comment-id:2153090251 --> @Zibri commented on GitHub (Jun 6, 2024): Hmm I think I found the reason. Creating a model as I did the template is not taken from the model but must be specified. So I added /set template [INST] {{ if .System }}{{ .System }} {{ end }}{{ .Prompt }} [/INST] And the system prompt worked. Now the problem I have is that it talks too much :D I wonder what paremeter limits the length of the responses without truncating them...
GiteaMirror commented 2026-04-12 13:30:59 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Jun 6, 2024):

@Zibri you can use the TEMPLATE command in the Modelfile to specify the template, and you'll need to also add some stop parameters. It should look something like:

FROM I:\models\Mistral-7b-Instruct-v0.3.f16.q6_k.gguf
TEMPLATE [INST] {{ .System }} {{ .Prompt }} [/INST]
PARAMETER stop [inst]
PARAMETER stop [/INST]

HTH. I'll close out the issue, but feel free to keep commenting.

<!-- gh-comment-id:2153434413 --> @pdevine commented on GitHub (Jun 6, 2024): @Zibri you can use the `TEMPLATE` command in the Modelfile to specify the template, and you'll need to also add some stop parameters. It should look something like: ``` FROM I:\models\Mistral-7b-Instruct-v0.3.f16.q6_k.gguf TEMPLATE [INST] {{ .System }} {{ .Prompt }} [/INST] PARAMETER stop [inst] PARAMETER stop [/INST] ``` HTH. I'll close out the issue, but feel free to keep commenting.
GiteaMirror commented 2026-04-12 13:30:59 -05:00
Author
Owner
Copy Link

@AditiMondal1110 commented on GitHub (Jul 17, 2025):

mistral-large through ollama iS not working,has anyone faced the same issue ? (I am using RTX4000ada onn runpod.)

<!-- gh-comment-id:3083489440 --> @AditiMondal1110 commented on GitHub (Jul 17, 2025): mistral-large through ollama iS not working,has anyone faced the same issue ? (I am using RTX4000ada onn runpod.)
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#3076
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?