github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-16 05:05:37 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #8421] tool_choice argument is being ignored #51921

New Issue
Closed
opened 2026-04-28 21:14:35 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-28 21:14:35 -05:00
Owner
Copy Link

Originally created by @pminervini on GitHub (Jan 14, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/8421

What is the issue?

Consider the following snippet:

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

from openai import OpenAI

client_kwargs = {
    'base_url': "http://localhost:11434/v1"
}

client = OpenAI(**client_kwargs)

# Messages from the log
messages = [{
    'role': 'system', 
    'content': 'You are a Kaggle grandmaster attending a competition.'
}]

# Tools specification from the log
tools = [{
    'type': 'function',
    'function': {
        'name': 'submit_review',
        'description': 'Submit a review evaluating the output of the training script.',
        'parameters': {
            'type': 'object',
            'properties': {
                'is_bug': {
                    'type': 'boolean',
                    'description': 'true if the output log shows that the execution failed or has some bug, otherwise false.'
                },
                'summary': {
                    'type': 'string',
                    'description': 'if there is a bug, propose a fix. Otherwise, write a short summary (2-3 sentences) describing the empirical findings.'
                },
                'metric': {
                    'type': 'number',
                    'description': 'If the code ran successfully, report the value of the validation metric. Otherwise, leave it null.'
                },
                'lower_is_better': {
                    'type': 'boolean',
                    'description': 'true if the metric should be minimized (i.e. a lower metric value is better, such as with MSE), false if the metric should be maximized (i.e. a higher metric value is better, such as with accuracy).'
                }
            },
            'required': ['is_bug', 'summary', 'metric', 'lower_is_better']
        }
    }
}]

# Make the API call with exact kwargs from the log
response = client.chat.completions.create(
    model='llama3.3' if 'base_url' in client_kwargs else 'gpt-4o',
    temperature=0.5,
    messages=messages,
    tools=tools,
    tool_choice={
        'type': 'function',
        'function': {'name': 'submit_review'}
    }
)

print(response)

tool_calls = response.choices[0].message.tool_calls

# Print the function call details
print(f"Tool calls: {tool_calls}")

Output:

ChatCompletion(id='chatcmpl-651', choices=[Choice(finish_reason='stop', index=0, logprobs=None, message=ChatCompletionMessage(content='', refusal=None, role='assistant', audio=None, function_call=None, tool_calls=None))], created=1736861285, model='llama3.3', object='chat.completion', service_tier=None, system_fingerprint='fp_ollama', usage=CompletionUsage(completion_tokens=1, prompt_tokens=59, total_tokens=60, completion_tokens_details=None, prompt_tokens_details=None))
Tool calls: None

Otoh, when using gpt-4o, I get the following:

ChatCompletion(id='chatcmpl-ApbGxne0uBHdxD8h44XhqCMX7Er3h', choices=[Choice(finish_reason='stop', index=0, logprobs=None, message=ChatCompletionMessage(content=None, refusal=None, role='assistant', audio=None, function_call=None, tool_calls=[ChatCompletionMessageToolCall(id='call_DWjp36UQSwc6r8UYDZB4kIct', function=Function(arguments='{"is_bug":false,"summary":"The training script executed successfully, achieving a validation metric of 0.85. This indicates a strong performance, suggesting that the model is well-tuned for the task at hand.","metric":0.85,"lower_is_better":false}', name='submit_review'), type='function')]))], created=1736861367, model='gpt-4o-2024-08-06', object='chat.completion', service_tier='default', system_fingerprint='fp_4691090a87', usage=CompletionUsage(completion_tokens=58, prompt_tokens=194, total_tokens=252, completion_tokens_details=CompletionTokensDetails(accepted_prediction_tokens=0, audio_tokens=0, reasoning_tokens=0, rejected_prediction_tokens=0), prompt_tokens_details=PromptTokensDetails(audio_tokens=0, cached_tokens=0)))
Tool calls: [ChatCompletionMessageToolCall(id='call_DWjp36UQSwc6r8UYDZB4kIct', function=Function(arguments='{"is_bug":false,"summary":"The training script executed successfully, achieving a validation metric of 0.85. This indicates a strong performance, suggesting that the model is well-tuned for the task at hand.","metric":0.85,"lower_is_better":false}', name='submit_review'), type='function')]

OS

macOS

GPU

Apple

CPU

Apple

Ollama version

0.5.5

Originally created by @pminervini on GitHub (Jan 14, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/8421 ### What is the issue? Consider the following snippet: ```python #!/usr/bin/env python3 # -*- coding: utf-8 -*- from openai import OpenAI client_kwargs = { 'base_url': "http://localhost:11434/v1" } client = OpenAI(**client_kwargs) # Messages from the log messages = [{ 'role': 'system', 'content': 'You are a Kaggle grandmaster attending a competition.' }] # Tools specification from the log tools = [{ 'type': 'function', 'function': { 'name': 'submit_review', 'description': 'Submit a review evaluating the output of the training script.', 'parameters': { 'type': 'object', 'properties': { 'is_bug': { 'type': 'boolean', 'description': 'true if the output log shows that the execution failed or has some bug, otherwise false.' }, 'summary': { 'type': 'string', 'description': 'if there is a bug, propose a fix. Otherwise, write a short summary (2-3 sentences) describing the empirical findings.' }, 'metric': { 'type': 'number', 'description': 'If the code ran successfully, report the value of the validation metric. Otherwise, leave it null.' }, 'lower_is_better': { 'type': 'boolean', 'description': 'true if the metric should be minimized (i.e. a lower metric value is better, such as with MSE), false if the metric should be maximized (i.e. a higher metric value is better, such as with accuracy).' } }, 'required': ['is_bug', 'summary', 'metric', 'lower_is_better'] } } }] # Make the API call with exact kwargs from the log response = client.chat.completions.create( model='llama3.3' if 'base_url' in client_kwargs else 'gpt-4o', temperature=0.5, messages=messages, tools=tools, tool_choice={ 'type': 'function', 'function': {'name': 'submit_review'} } ) print(response) tool_calls = response.choices[0].message.tool_calls # Print the function call details print(f"Tool calls: {tool_calls}") ``` Output: ```bash ChatCompletion(id='chatcmpl-651', choices=[Choice(finish_reason='stop', index=0, logprobs=None, message=ChatCompletionMessage(content='', refusal=None, role='assistant', audio=None, function_call=None, tool_calls=None))], created=1736861285, model='llama3.3', object='chat.completion', service_tier=None, system_fingerprint='fp_ollama', usage=CompletionUsage(completion_tokens=1, prompt_tokens=59, total_tokens=60, completion_tokens_details=None, prompt_tokens_details=None)) Tool calls: None ``` Otoh, when using `gpt-4o`, I get the following: ```bash ChatCompletion(id='chatcmpl-ApbGxne0uBHdxD8h44XhqCMX7Er3h', choices=[Choice(finish_reason='stop', index=0, logprobs=None, message=ChatCompletionMessage(content=None, refusal=None, role='assistant', audio=None, function_call=None, tool_calls=[ChatCompletionMessageToolCall(id='call_DWjp36UQSwc6r8UYDZB4kIct', function=Function(arguments='{"is_bug":false,"summary":"The training script executed successfully, achieving a validation metric of 0.85. This indicates a strong performance, suggesting that the model is well-tuned for the task at hand.","metric":0.85,"lower_is_better":false}', name='submit_review'), type='function')]))], created=1736861367, model='gpt-4o-2024-08-06', object='chat.completion', service_tier='default', system_fingerprint='fp_4691090a87', usage=CompletionUsage(completion_tokens=58, prompt_tokens=194, total_tokens=252, completion_tokens_details=CompletionTokensDetails(accepted_prediction_tokens=0, audio_tokens=0, reasoning_tokens=0, rejected_prediction_tokens=0), prompt_tokens_details=PromptTokensDetails(audio_tokens=0, cached_tokens=0))) Tool calls: [ChatCompletionMessageToolCall(id='call_DWjp36UQSwc6r8UYDZB4kIct', function=Function(arguments='{"is_bug":false,"summary":"The training script executed successfully, achieving a validation metric of 0.85. This indicates a strong performance, suggesting that the model is well-tuned for the task at hand.","metric":0.85,"lower_is_better":false}', name='submit_review'), type='function')] ``` ### OS macOS ### GPU Apple ### CPU Apple ### Ollama version 0.5.5
GiteaMirror added the bug label 2026-04-28 21:14:35 -05:00
GiteaMirror commented 2026-04-28 21:14:37 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Jan 15, 2025):

https://github.com/ollama/ollama/blob/main/docs/openai.md#supported-request-fields:~:text=tools-,tool_choice,-logit_bias

<!-- gh-comment-id:2591370676 --> @rick-github commented on GitHub (Jan 15, 2025): https://github.com/ollama/ollama/blob/main/docs/openai.md#supported-request-fields:~:text=tools-,tool_choice,-logit_bias
GiteaMirror commented 2026-04-28 21:14:38 -05:00
Author
Owner
Copy Link

@pminervini commented on GitHub (Jan 15, 2025):

Oof, thank you! Closing (& looking into how to add this feature)

<!-- gh-comment-id:2591876892 --> @pminervini commented on GitHub (Jan 15, 2025): Oof, thank you! Closing (& looking into how to add this feature)
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#51921
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?