github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 16:11:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #3151] Doubt about openai compatibility with temperature parameter #48449

New Issue
Closed
opened 2026-04-28 08:24:27 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-28 08:24:27 -05:00
Owner
Copy Link

Originally created by @ejgutierrez74 on GitHub (Mar 14, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/3151

I have one doubt, about the use of temperature.
As i have understood, temperature in llama2 is from 0.0 to 1.0.

But if you use chat.completion from openai ( https://github.com/ollama/ollama/blob/main/docs/openai.md), if you read the documentation temperature values range from 0.0 to 2.0, so seems a little mismatch.

For example if you write temperature 1.0, in llama2 would be the more random parameter, but in case of openai, would be medium ( the max should be 2.0)

So when i use:


chat_completion = client.chat.completions.create(
    messages=[
        {
            'role': 'user',
            'content': 'Say this is a test',
        }
    ],
    model='llama2',
    temperature=1.0
)

What should i expect, medium randomness as openai documentation, or full random as llama2

Thanks in advance

Originally created by @ejgutierrez74 on GitHub (Mar 14, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/3151 I have one doubt, about the use of temperature. As i have understood, temperature in llama2 is from 0.0 to 1.0. But if you use chat.completion from openai ( https://github.com/ollama/ollama/blob/main/docs/openai.md), if you read the documentation temperature values range from 0.0 to 2.0, so seems a little mismatch. For example if you write temperature 1.0, in llama2 would be the more random parameter, but in case of openai, would be medium ( the max should be 2.0) So when i use: ``` chat_completion = client.chat.completions.create( messages=[ { 'role': 'user', 'content': 'Say this is a test', } ], model='llama2', temperature=1.0 ) ``` What should i expect, medium randomness as openai documentation, or full random as llama2 Thanks in advance
GiteaMirror commented 2026-04-28 08:24:29 -05:00
Author
Owner
Copy Link

@igorschlum commented on GitHub (Mar 14, 2024):

@ejgutierrez74 I think you have a good point. If you work with ollama using ChatGPT format, temperature should range from 0 to 2 and then Ollama should divide temperature by 2 before processing the prompt.

<!-- gh-comment-id:1998646407 --> @igorschlum commented on GitHub (Mar 14, 2024): @ejgutierrez74 I think you have a good point. If you work with ollama using ChatGPT format, temperature should range from 0 to 2 and then Ollama should divide temperature by 2 before processing the prompt.
GiteaMirror commented 2026-04-28 08:24:31 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Mar 15, 2024):

1.0 should be median – it's adjusted to a rough equivalent with llama2 and other models. Hope this helps, and let me know if you see anything off 😊

<!-- gh-comment-id:1998754915 --> @jmorganca commented on GitHub (Mar 15, 2024): 1.0 should be median – it's adjusted to a rough equivalent with llama2 and other models. Hope this helps, and let me know if you see anything off 😊
GiteaMirror commented 2026-04-28 08:24:33 -05:00
Author
Owner
Copy Link

@ejgutierrez74 commented on GitHub (Mar 15, 2024):

Im following a course of Llama2, seems official as the former is Amit Sangani Senior Director of Eng @Meta. It says that in Llama the value varies from 0 to 1 ( says that 0.9 is a higher value). Thats the doubt....id follow the course and try to get possible inconsistencies between openai official API and Llama2 LLM

Thanks
@jmorganca by the way i reopened the ubuntu bug with ollama
.

<!-- gh-comment-id:2000235667 --> @ejgutierrez74 commented on GitHub (Mar 15, 2024): Im following a course of Llama2, seems official as the former is Amit Sangani Senior Director of Eng [@Meta](https://twitter.com/Meta). It says that in Llama the value varies from 0 to 1 ( says that 0.9 is a higher value). Thats the doubt....id follow the course and try to get possible inconsistencies between openai official API and Llama2 LLM Thanks @jmorganca by the way i reopened the ubuntu bug with ollama .
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
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#48449
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?