1  from mlflow.entities import Metric
  2  from mlflow.evaluation.assessment import AssessmentEntity, AssessmentSource
  3  from mlflow.evaluation.evaluation import EvaluationEntity
  4  from mlflow.evaluation.evaluation_tag import EvaluationTag
  5  
  6  
  7  def test_evaluation_equality():
  8      source_1 = AssessmentSource(source_type="HUMAN", source_id="user_1")
  9      metric_1 = Metric(key="metric1", value=1.1, timestamp=123, step=0)
 10      tag_1 = EvaluationTag(key="tag1", value="value1")
 11  
 12      # Valid evaluations
 13      evaluation_1 = EvaluationEntity(
 14          evaluation_id="eval1",
 15          run_id="run1",
 16          inputs_id="inputs1",
 17          inputs={"feature1": 1.0, "feature2": 2.0},
 18          outputs={"prediction": 0.5},
 19          request_id="req1",
 20          targets={"actual": 0.6},
 21          assessments=[
 22              AssessmentEntity(
 23                  evaluation_id="eval1",
 24                  name="relevance",
 25                  source=source_1,
 26                  timestamp=123456789,
 27                  numeric_value=0.9,
 28              )
 29          ],
 30          metrics=[metric_1],
 31          tags=[tag_1],
 32          error_code="E001",
 33          error_message="An error occurred",
 34      )
 35      evaluation_2 = EvaluationEntity(
 36          evaluation_id="eval1",
 37          run_id="run1",
 38          inputs_id="inputs1",
 39          inputs={"feature1": 1.0, "feature2": 2.0},
 40          outputs={"prediction": 0.5},
 41          request_id="req1",
 42          targets={"actual": 0.6},
 43          assessments=[
 44              AssessmentEntity(
 45                  evaluation_id="eval1",
 46                  name="relevance",
 47                  source=source_1,
 48                  timestamp=123456789,
 49                  numeric_value=0.9,
 50              )
 51          ],
 52          metrics=[metric_1],
 53          tags=[tag_1],
 54          error_code="E001",
 55          error_message="An error occurred",
 56      )
 57      evaluation_3 = EvaluationEntity(
 58          evaluation_id="eval2",
 59          run_id="run2",
 60          inputs_id="inputs2",
 61          inputs={"feature1": 1.0, "feature2": 2.0},
 62          outputs={"prediction": 0.5},
 63          request_id="req2",
 64          targets={"actual": 0.7},
 65          assessments=[
 66              AssessmentEntity(
 67                  evaluation_id="eval2",
 68                  name="relevance",
 69                  source=source_1,
 70                  timestamp=123456789,
 71                  numeric_value=0.8,
 72              )
 73          ],
 74          metrics=[Metric(key="metric1", value=1.2, timestamp=123, step=0)],
 75          tags=[EvaluationTag(key="tag2", value="value2")],
 76          error_code="E002",
 77          error_message="Another error occurred",
 78      )
 79  
 80      assert evaluation_1 == evaluation_2  # Same evaluation data
 81      assert evaluation_1 != evaluation_3  # Different evaluation data
 82  
 83  
 84  def test_evaluation_properties():
 85      source = AssessmentSource(source_type="HUMAN", source_id="user_1")
 86      metric = Metric(key="metric1", value=1.1, timestamp=123, step=0)
 87      tag = EvaluationTag(key="tag1", value="value1")
 88      assessment = AssessmentEntity(
 89          evaluation_id="eval1",
 90          name="relevance",
 91          source=source,
 92          timestamp=123456789,
 93          numeric_value=0.9,
 94          rationale="Rationale text",
 95          metadata={"key1": "value1"},
 96      )
 97      evaluation = EvaluationEntity(
 98          evaluation_id="eval1",
 99          run_id="run1",
100          inputs_id="inputs1",
101          inputs={"feature1": 1.0, "feature2": 2.0},
102          outputs={"prediction": 0.5},
103          request_id="req1",
104          targets={"actual": 0.6},
105          assessments=[assessment],
106          metrics=[metric],
107          tags=[tag],
108          error_code="E001",
109          error_message="An error occurred",
110      )
111  
112      assert evaluation.evaluation_id == "eval1"
113      assert evaluation.run_id == "run1"
114      assert evaluation.inputs_id == "inputs1"
115      assert evaluation.inputs == {"feature1": 1.0, "feature2": 2.0}
116      assert evaluation.outputs == {"prediction": 0.5}
117      assert evaluation.request_id == "req1"
118      assert evaluation.targets == {"actual": 0.6}
119      assert evaluation.error_code == "E001"
120      assert evaluation.error_message == "An error occurred"
121      assert evaluation.assessments == [assessment]
122      assert evaluation.metrics == [metric]
123      assert evaluation.tags == [tag]
124  
125  
126  def test_evaluation_to_from_dictionary():
127      source = AssessmentSource(source_type="HUMAN", source_id="user_1")
128      metric = Metric(key="metric1", value=1.1, timestamp=123, step=0)
129      tag = EvaluationTag(key="tag1", value="value1")
130      assessment = AssessmentEntity(
131          evaluation_id="eval1",
132          name="relevance",
133          source=source,
134          timestamp=123456789,
135          numeric_value=0.9,
136          rationale="Rationale text",
137          metadata={"key1": "value1"},
138      )
139      evaluation = EvaluationEntity(
140          evaluation_id="eval1",
141          run_id="run1",
142          inputs_id="inputs1",
143          inputs={"feature1": 1.0, "feature2": 2.0},
144          outputs={"prediction": 0.5},
145          request_id="req1",
146          targets={"actual": 0.6},
147          assessments=[assessment],
148          metrics=[metric],
149          tags=[tag],
150          error_code="E001",
151          error_message="An error occurred",
152      )
153      evaluation_dict = evaluation.to_dictionary()
154  
155      expected_dict = {
156          "evaluation_id": "eval1",
157          "run_id": "run1",
158          "inputs_id": "inputs1",
159          "inputs": {"feature1": 1.0, "feature2": 2.0},
160          "outputs": {"prediction": 0.5},
161          "request_id": "req1",
162          "targets": {"actual": 0.6},
163          "assessments": [assessment.to_dictionary()],
164          "metrics": [metric.to_dictionary()],
165          "tags": [tag.to_dictionary()],
166          "error_code": "E001",
167          "error_message": "An error occurred",
168      }
169      assert evaluation_dict == expected_dict
170  
171      recreated_evaluation = EvaluationEntity.from_dictionary(evaluation_dict)
172      assert recreated_evaluation == evaluation
173  
174  
175  def test_evaluation_construction_with_minimal_required_fields():
176      evaluation = EvaluationEntity(
177          evaluation_id="eval1",
178          run_id="run1",
179          inputs_id="inputs1",
180          inputs={"feature1": 1.0, "feature2": 2.0},
181      )
182      evaluation_dict = evaluation.to_dictionary()
183      recreated_evaluation = EvaluationEntity.from_dictionary(evaluation_dict)
184      assert recreated_evaluation == evaluation