Skip to content

[evaluation] Enable groundedness for Bing and OpenAPI tool calls - #48304

Closed
mmkawale wants to merge 1 commit into
Azure:mainfrom
mmkawale:copilot/groundedness-enable-bing-openapi-sdk-mainline-refresh
Closed

[evaluation] Enable groundedness for Bing and OpenAPI tool calls#48304
mmkawale wants to merge 1 commit into
Azure:mainfrom
mmkawale:copilot/groundedness-enable-bing-openapi-sdk-mainline-refresh

Conversation

@mmkawale

Copy link
Copy Markdown
Contributor

Summary

  • enable groundedness validator support for:
    • bing_grounding
    • bing_custom_search
    • openapi_call
  • keep support for already-enabled structured tools:
    • azure_ai_search
    • azure_fabric
    • sharepoint_grounding
  • keep guardrails for still-unsupported tools:
    • browser_automation
    • code_interpreter_call
    • computer_call
    • web_search

Changes

  • updated GroundednessConversationValidator.UNSUPPORTED_TOOLS to align with current groundedness policy
  • updated groundedness evaluator comments to match new validator behavior
  • updated validator unit tests in test_common_validators.py to validate the new list and behavior

Validation

  • python -m black on touched files
  • PYTHONPATH=sdk/evaluation/azure-ai-evaluation python -m pytest sdk/evaluation/azure-ai-evaluation/tests/unittests/test_common_validators.py -k "UnsupportedToolsListGroundednessValidator or UnsupportedToolsListConversationValidator or ConversationValidatorRejectsStillUnsupportedTools"
  • result: 24 passed

Align GroundednessConversationValidator with assets policy: allow bing_grounding, bing_custom_search, and openapi_call in addition to already-enabled azure_ai_search/azure_fabric/sharepoint_grounding; retain guardrails for browser_automation/code_interpreter_call/computer_call/web_search.

Update groundedness validator comments and validator unit tests to reflect the narrowed unsupported list.

Validation: black on touched files; PYTHONPATH=sdk/evaluation/azure-ai-evaluation pytest tests/unittests/test_common_validators.py -k 'UnsupportedToolsListGroundednessValidator or UnsupportedToolsListConversationValidator or ConversationValidatorRejectsStillUnsupportedTools' (24 passed).
@mmkawale
mmkawale requested a review from a team as a code owner July 28, 2026 19:01
Copilot AI review requested due to automatic review settings July 28, 2026 19:01
@github-actions github-actions Bot added the Evaluation Issues related to the client library for Azure AI Evaluation label Jul 28, 2026
@azure-pipelines

Copy link
Copy Markdown
Azure Pipelines:
Successfully started running 1 pipeline(s).
9 pipeline(s) were filtered out due to trigger conditions.
There may be pipelines that require an authorized user to comment /azp run to run.

Copilot AI left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

Enables additional groundedness tool-call validation while retaining guardrails for unsupported tools.

Changes:

  • Narrows Groundedness’s unsupported-tool list.
  • Updates validator tests and evaluator comments.
  • Applies Black formatting to touched code.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 3 comments.

File Description
_conversation_validator.py Revises Groundedness tool restrictions.
_groundedness.py Updates evaluator comments and formatting.
test_common_validators.py Updates supported-tool validation tests.

Comment on lines +531 to +534
UNSUPPORTED_TOOLS: List[str] = [
tool_name
for tool_name in ConversationValidator.UNSUPPORTED_TOOLS
if tool_name not in {"bing_custom_search", "bing_grounding", "openapi_call"}
Comment on lines +123 to +125
# keeps stricter guardrails than the shared base for the remaining
# unsupported tool families, while allowing grounding/search/openapi
# tool calls supported by this evaluator.
Comment on lines +761 to +764
@pytest.mark.parametrize(
"tool_name", NEWLY_ENABLED_TOOLS + GROUNDEDNESS_NEWLY_ENABLED_TOOLS
)
def test_groundedness_validator_accepts_newly_enabled_tools(self, tool_name):
@mmkawale

Copy link
Copy Markdown
Contributor Author

Superseded by #48305, which contains the same groundedness behavior updates with significantly reduced whitespace/formatting churn for easier review.

@mmkawale

Copy link
Copy Markdown
Contributor Author

Closing in favor of #48305 (clean replacement with minimal diff noise).

@mmkawale mmkawale closed this Jul 28, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Evaluation Issues related to the client library for Azure AI Evaluation

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants