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 #2844] OpenAI package compatibility #1730

New Issue
Closed
opened 2026-04-12 11:42:39 -05:00 by GiteaMirror · 6 comments
GiteaMirror commented 2026-04-12 11:42:39 -05:00
Owner
Copy Link

Originally created by @eliranwong on GitHub (Feb 29, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/2844

I read the example in https://ollama.com/blog/openai-compatibility

The example works, but it doesn't when I add "response_format={ "type": "json_object" },"

https://platform.openai.com/docs/guides/text-generation/json-mode

Originally created by @eliranwong on GitHub (Feb 29, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/2844 I read the example in https://ollama.com/blog/openai-compatibility The example works, but it doesn't when I add "response_format={ "type": "json_object" }," https://platform.openai.com/docs/guides/text-generation/json-mode
GiteaMirror commented 2026-04-12 11:42:40 -05:00
Author
Owner
Copy Link

@eliranwong commented on GitHub (Feb 29, 2024):

ok, I just found a workaround, by adding the following prefix in User message content:

Use template '{"answer": ""}' to answer:

<!-- gh-comment-id:1971993292 --> @eliranwong commented on GitHub (Feb 29, 2024): ok, I just found a workaround, by adding the following prefix in User message content: Use template '{"answer": ""}' to answer:
GiteaMirror commented 2026-04-12 11:42:40 -05:00
Author
Owner
Copy Link

@eliranwong commented on GitHub (Feb 29, 2024):

I am happy enough with the workaround. many thanks.

<!-- gh-comment-id:1971994264 --> @eliranwong commented on GitHub (Feb 29, 2024): I am happy enough with the workaround. many thanks.
GiteaMirror commented 2026-04-12 11:42:41 -05:00
Author
Owner
Copy Link

@eliranwong commented on GitHub (Feb 29, 2024):

By the way, the full workaround example that works is:

from openai import OpenAI
client = OpenAI( 
    base_url = 'http://localhost:11434/v1', 
    api_key='ollama', # required, but unused 
) 
 
response = client.chat.completions.create( 
  model="llama2", 
  response_format={ "type": "json_object" }, 
  messages=[ 
    {"role": "system", "content": "You are a helpful assistant."}, 
    {"role": "user", "content": "Who won the world series in 2020?"}, 
    {"role": "assistant", "content": "The LA Dodgers won in 2020."}, 
    {"role": "user", "content": """Use template '{"answer": ""}' to answer: Wher
e was it played?"""} 
  ] 
) 
print(response.choices[0].message.content)
<!-- gh-comment-id:1971995760 --> @eliranwong commented on GitHub (Feb 29, 2024): By the way, the full workaround example that works is: ``` from openai import OpenAI client = OpenAI( base_url = 'http://localhost:11434/v1', api_key='ollama', # required, but unused ) response = client.chat.completions.create( model="llama2", response_format={ "type": "json_object" }, messages=[ {"role": "system", "content": "You are a helpful assistant."}, {"role": "user", "content": "Who won the world series in 2020?"}, {"role": "assistant", "content": "The LA Dodgers won in 2020."}, {"role": "user", "content": """Use template '{"answer": ""}' to answer: Wher e was it played?"""} ] ) print(response.choices[0].message.content) ```
GiteaMirror commented 2026-04-12 11:42:42 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Mar 1, 2024):

Hi there, thanks for opening an issue. If using json_object, it's very important to tell the model to answer in JSON, or it may get stuck generating a lot of whitespace (solving this separately with #2605). The OpenAI API behaves similarly although it will error if "answer in JSON" or similar isn't in the prompt. Let me know if you still see more issues!

<!-- gh-comment-id:1972260932 --> @jmorganca commented on GitHub (Mar 1, 2024): Hi there, thanks for opening an issue. If using `json_object`, it's very important to tell the model to answer in JSON, or it may get stuck generating a lot of whitespace (solving this separately with #2605). The OpenAI API behaves similarly although it will error if "answer in JSON" or similar isn't in the prompt. Let me know if you still see more issues!
GiteaMirror commented 2026-04-12 11:42:43 -05:00
Author
Owner
Copy Link

@halcwb commented on GitHub (Mar 13, 2024):

@jmorganca Tried the official way with a response_format object with type=json_object and schema filled in with a schema object, but that is ignored. When I try to put the schema in the prompt then I get the white space problem.

Using the exact same code with for example the openAI api from fireworks, it works fine.

<!-- gh-comment-id:1994261460 --> @halcwb commented on GitHub (Mar 13, 2024): @jmorganca Tried the official way with a response_format object with type=json_object and schema filled in with a schema object, but that is ignored. When I try to put the schema in the prompt then I get the white space problem. Using the exact same code with for example the openAI api from fireworks, it works fine.
GiteaMirror commented 2026-04-12 11:42:43 -05:00
Author
Owner
Copy Link

@PhilipAmadasun commented on GitHub (May 2, 2024):

@jmorganca I'll create an issue for this but at this time is it possible for developers to create API keys for access security using the openai API (with ollama for inference)?

<!-- gh-comment-id:2089375627 --> @PhilipAmadasun commented on GitHub (May 2, 2024): @jmorganca I'll create an issue for this but at this time is it possible for developers to create API keys for access security using the openai API (with ollama for inference)?
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#1730
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?