1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
|
from unittest.mock import MagicMock, patch
import pytest
from azure.ai.evaluation._exceptions import EvaluationException
from azure.ai.evaluation import (
FluencyEvaluator,
SimilarityEvaluator,
RetrievalEvaluator,
RelevanceEvaluator,
GroundednessEvaluator,
)
async def quality_response_async_mock(*args, **kwargs):
llm_output = (
"<S0>Let's think step by step: The response 'Honolulu' is a single word. "
"It does not form a complete sentence, lacks grammatical structure, and does not "
"convey any clear idea or message. It is not possible to assess vocabulary range, "
"sentence complexity, coherence, or overall readability from a single word. Therefore,"
"it falls into the category of minimal command of the language.</S0>"
"<S1>The response is a single word and does not provide any meaningful content to evaluate"
" fluency. It is largely incomprehensible and does not meet the criteria for higher fluency "
"levels.</S1><S2>1</S2>"
)
return {"llm_output": llm_output}
async def quality_no_response_async_mock():
return {"llm_output": "1"}
@pytest.mark.usefixtures("mock_model_config")
@pytest.mark.unittest
class TestBuiltInEvaluators:
def test_fluency_evaluator(self, mock_model_config):
fluency_eval = FluencyEvaluator(model_config=mock_model_config)
fluency_eval._flow = MagicMock(return_value=quality_response_async_mock())
score = fluency_eval(response="The capital of Japan is Tokyo.")
assert score is not None
assert score["fluency"] == score["gpt_fluency"] == 1
def test_fluency_evaluator_non_string_inputs(self, mock_model_config):
fluency_eval = FluencyEvaluator(model_config=mock_model_config)
fluency_eval._flow = MagicMock(return_value=quality_response_async_mock())
score = fluency_eval(response={"bar": "2"})
assert score is not None
assert score["fluency"] == score["gpt_fluency"] == 1
def test_fluency_evaluator_empty_string(self, mock_model_config):
fluency_eval = FluencyEvaluator(model_config=mock_model_config)
fluency_eval._flow = MagicMock(return_value=quality_response_async_mock())
with pytest.raises(EvaluationException) as exc_info:
fluency_eval(response=None)
assert (
"FluencyEvaluator: Either 'conversation' or individual inputs must be provided." in exc_info.value.args[0]
)
def test_similarity_evaluator_keys(self, mock_model_config):
similarity_eval = SimilarityEvaluator(model_config=mock_model_config)
similarity_eval._flow = MagicMock(return_value=quality_no_response_async_mock())
result = similarity_eval(
query="What is the capital of Japan?",
response="The capital of Japan is Tokyo.",
ground_truth="Tokyo is Japan's capital, known for its blend of traditional culture and technological advancements.",
)
assert result["similarity"] == result["gpt_similarity"] == 1
# Updated assertion to expect 4 keys instead of 2
assert len(result) == 11
# Verify all expected keys are present
assert set(result.keys()) == {
"similarity",
"gpt_similarity",
"similarity_result",
"similarity_threshold",
"similarity_prompt_tokens",
"similarity_completion_tokens",
"similarity_total_tokens",
"similarity_finish_reason",
"similarity_model",
"similarity_sample_input",
"similarity_sample_output",
}
def test_retrieval_evaluator_keys(self, mock_model_config):
retrieval_eval = RetrievalEvaluator(model_config=mock_model_config)
retrieval_eval._flow = MagicMock(return_value=quality_response_async_mock())
result = retrieval_eval(
query="What is the value of 2 + 2?",
context="1 + 2 = 2",
)
assert result["retrieval"] == result["gpt_retrieval"] == 1
assert result["retrieval"] == result["gpt_retrieval"]
assert result["retrieval_reason"]
retrieval_eval = RetrievalEvaluator(model_config=mock_model_config)
retrieval_eval._flow = MagicMock(return_value=quality_response_async_mock())
conversation = {
"messages": [
{"role": "user", "content": "What is the value of 2 + 2?"},
{
"role": "assistant",
"content": "2 + 2 = 4",
"context": {
"citations": [
{"id": "math_doc.md", "content": "Information about additions: 1 + 2 = 3, 2 + 2 = 4"}
]
},
},
]
}
result = retrieval_eval(conversation=conversation)
assert result["retrieval"] == result["gpt_retrieval"] == 1
retrieval_eval = RetrievalEvaluator(model_config=mock_model_config)
retrieval_eval._flow = MagicMock(return_value=quality_response_async_mock())
conversation = {
"messages": [
{"role": "user", "content": "What is the value of 2 + 2?"},
{
"role": "assistant",
"content": "2 + 2 = 4",
"context": "Information about additions: 1 + 2 = 3, 2 + 2 = 4",
},
]
}
result = retrieval_eval(conversation=conversation)
assert result["retrieval"] == result["gpt_retrieval"] == 1
def test_quality_evaluator_missing_input(self, mock_model_config):
"""All evaluators that inherit from EvaluatorBase are covered by this test"""
quality_eval = RetrievalEvaluator(model_config=mock_model_config)
quality_eval._flow = MagicMock(return_value=quality_response_async_mock())
with pytest.raises(EvaluationException) as exc_info:
quality_eval(response="The capital of Japan is Tokyo.") # Retrieval requires query and context
assert (
"RetrievalEvaluator: Either 'conversation' or individual inputs must be provided." in exc_info.value.args[0]
)
@patch("azure.ai.evaluation._evaluators._groundedness._groundedness.AsyncPrompty.load")
def test_groundedness_evaluator_with_agent_response(self, mock_async_prompty, mock_model_config):
"""Test GroundednessEvaluator with query, response, and tool_definitions"""
groundedness_eval = GroundednessEvaluator(model_config=mock_model_config)
mock_async_prompty.return_value = quality_response_async_mock
# Test with query, response, and tool_definitions
result = groundedness_eval(
query="What is the capital of Japan?",
response=[
{
"createdAt": "2025-08-01T00:02:38Z",
"run_id": "run_CmSdDdrq0CzwGOwqmWVADYwi",
"tool_call_id": "call_AU6kCcVwxv1cjM8HIQHMFFGh",
"role": "tool",
"content": [
{
"type": "tool_result",
"tool_result": [
{
"file_id": "assistant-6QeBNfMsJpL3AHnE3T6dwY",
"file_name": "product_info_1.md",
"score": 0.03333333507180214,
"attributes": {},
"content": [
{
"type": "text",
"text": "# Information about product item_number: 1\n\n## Brand\nContoso Galaxy Innovations\n\n## Category\nSmart Eyewear\n",
}
],
}
],
}
],
},
{
"createdAt": "2025-08-01T00:02:38Z",
"run_id": "run_CmSdDdrq0CzwGOwqmWVADYwi",
"role": "assistant",
"content": [
{"type": "text", "text": "One of the Contoso products identified is the **SmartView Glasses**"}
],
},
{
"createdAt": "2025-08-01T00:02:38Z",
"run_id": "run_CmSdDdrq0CzwGOwqmWVADYwi",
"role": "assistant",
"content": [
{
"type": "tool_call",
"tool_call_id": "call_AU6kCcVwxv1cjM8HIQHMFFGh",
"name": "file_search",
"arguments": {"ranking_options": {"ranker": "default_2024_08_21", "score_threshold": 0.0}},
}
],
},
],
tool_definitions=[
{"name": "file_search", "type": "file_search", "description": "Search for information in files"}
],
)
assert result is not None
assert result["groundedness"] == result["gpt_groundedness"] == 1
assert "groundedness_reason" in result
def test_groundedness_evaluator_with_context(self, mock_model_config):
"""Test GroundednessEvaluator with direct context (traditional use)"""
groundedness_eval = GroundednessEvaluator(model_config=mock_model_config)
groundedness_eval._flow = MagicMock(return_value=quality_response_async_mock())
result = groundedness_eval(
response="The capital of Japan is Tokyo.",
context="Tokyo is the capital of Japan and is located on the eastern coast of Honshu island.",
)
assert result is not None
assert result["groundedness"] == result["gpt_groundedness"] == 1
assert "groundedness_reason" in result
def test_groundedness_evaluator_missing_required_inputs(self, mock_model_config):
"""Test GroundednessEvaluator with missing required inputs for agent response mode"""
groundedness_eval = GroundednessEvaluator(model_config=mock_model_config)
groundedness_eval._flow = MagicMock(return_value=quality_response_async_mock())
with pytest.raises(EvaluationException) as exc_info:
groundedness_eval(
query="What is the capital of Japan?",
# Missing response
)
assert (
"Either 'conversation' or individual inputs must be provided. For Agent groundedness 'query' and 'response' are required."
in exc_info.value.args[0]
)
|
