Skip to content

fix: GenAI Client(evals) - Update retry mechanism #6175

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
copybara-service wants to merge 1 commit into
base: main
Choose a base branch
from
Open

fix: GenAI Client(evals) - Update retry mechanism #6175

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
148 changes: 146 additions & 2 deletions tests/unit/vertexai/genai/test_evals.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1247,6 +1247,143 @@ def test_run_inference_with_agent_engine_with_response_column_raises_error(
"'intermediate_events' or 'response' columns"
) in str(excinfo.value)

@mock.patch.object(_evals_utils, "EvalDatasetLoader")
@mock.patch(
"vertexai._genai._evals_common.InMemorySessionService"
)
@mock.patch("vertexai._genai._evals_common.Runner")
@mock.patch("vertexai._genai._evals_common.LlmAgent")
def test_run_inference_with_local_agent(
self,
mock_llm_agent,
mock_runner,
mock_session_service,
mock_eval_dataset_loader,
):
mock_df = pd.DataFrame(
{
"prompt": ["agent prompt", "agent prompt 2"],
"session_inputs": [
{
"user_id": "123",
"state": {"a": "1"},
},
{
"user_id": "456",
"state": {"b": "2"},
},
],
}
)
mock_eval_dataset_loader.return_value.load.return_value = mock_df.to_dict(
orient="records"
)

mock_agent_instance = mock.Mock()
mock_llm_agent.return_value = mock_agent_instance
mock_session_service.return_value.create_session = mock.AsyncMock()
mock_runner_instance = mock_runner.return_value
stream_run_return_value_1 = [
mock.Mock(
model_dump=lambda: {
"id": "1",
"content": {"parts": [{"text": "intermediate1"}]},
"timestamp": 123,
"author": "model",
}
),
mock.Mock(
model_dump=lambda: {
"id": "2",
"content": {"parts": [{"text": "agent response"}]},
"timestamp": 124,
"author": "model",
}
),
]
stream_run_return_value_2 = [
mock.Mock(
model_dump=lambda: {
"id": "3",
"content": {"parts": [{"text": "intermediate2"}]},
"timestamp": 125,
"author": "model",
}
),
mock.Mock(
model_dump=lambda: {
"id": "4",
"content": {"parts": [{"text": "agent response 2"}]},
"timestamp": 126,
"author": "model",
}
),
]

async def async_iterator(items):
for item in items:
yield item

mock_runner_instance.run_async.side_effect = [
async_iterator(stream_run_return_value_1),
async_iterator(stream_run_return_value_2),
]

inference_result = self.client.evals.run_inference(
agent=mock_agent_instance,
src=mock_df,
)

mock_eval_dataset_loader.return_value.load.assert_called_once_with(mock_df)
assert mock_session_service.call_count == 2
mock_runner.assert_called_with(
agent=mock_agent_instance,
app_name="local agent run",
session_service=mock_session_service.return_value,
)
assert mock_runner.call_count == 2
assert mock_runner_instance.run_async.call_count == 2

pd.testing.assert_frame_equal(
inference_result.eval_dataset_df,
pd.DataFrame(
{
"prompt": ["agent prompt", "agent prompt 2"],
"session_inputs": [
{
"user_id": "123",
"state": {"a": "1"},
},
{
"user_id": "456",
"state": {"b": "2"},
},
],
"intermediate_events": [
[
{
"event_id": "1",
"content": {"parts": [{"text": "intermediate1"}]},
"creation_timestamp": 123,
"author": "model",
}
],
[
{
"event_id": "3",
"content": {"parts": [{"text": "intermediate2"}]},
"creation_timestamp": 125,
"author": "model",
}
],
],
"response": ["agent response", "agent response 2"],
}
),
)
assert inference_result.candidate_name is None
assert inference_result.gcs_source is None

def test_run_inference_with_litellm_string_prompt_format(
self,
mock_api_client_fixture,
Expand Down Expand Up @@ -1599,6 +1736,7 @@ def test_run_agent_internal_success(self, mock_run_agent):
result_df = _evals_common._run_agent_internal(
api_client=mock_api_client,
agent_engine=mock_agent_engine,
agent=None,
prompt_dataset=prompt_dataset,
)

Expand Down Expand Up @@ -1629,6 +1767,7 @@ def test_run_agent_internal_error_response(self, mock_run_agent):
result_df = _evals_common._run_agent_internal(
api_client=mock_api_client,
agent_engine=mock_agent_engine,
agent=None,
prompt_dataset=prompt_dataset,
)

Expand All @@ -1655,6 +1794,7 @@ def test_run_agent_internal_malformed_event(self, mock_run_agent):
result_df = _evals_common._run_agent_internal(
api_client=mock_api_client,
agent_engine=mock_agent_engine,
agent=None,
prompt_dataset=prompt_dataset,
)
assert "response" in result_df.columns
Expand Down Expand Up @@ -4984,7 +5124,9 @@ def test_execute_evaluation_adds_creation_timestamp(
frozenset(["summarization_quality"]),
)
@mock.patch("time.sleep", return_value=None)
@mock.patch("vertexai._genai.evals.Evals._evaluate_instances")
@mock.patch(
"vertexai._genai.evals.Evals._evaluate_instances"
)
def test_predefined_metric_retry_on_resource_exhausted(
self,
mock_private_evaluate_instances,
Expand Down Expand Up @@ -5037,7 +5179,9 @@ def test_predefined_metric_retry_on_resource_exhausted(
frozenset(["summarization_quality"]),
)
@mock.patch("time.sleep", return_value=None)
@mock.patch("vertexai._genai.evals.Evals._evaluate_instances")
@mock.patch(
"vertexai._genai.evals.Evals._evaluate_instances"
)
def test_predefined_metric_retry_fail_on_resource_exhausted(
self,
mock_private_evaluate_instances,
Expand Down
Loading
Loading