chore: import upstream snapshot with attribution

examples/flows/evaluation/eval-accuracy-maths-to-code/aggregate.py

@@ -0,0 +1,37 @@
+from typing import List
+from promptflow.core import tool
+from promptflow.core import log_metric
+
+
+@tool
+def accuracy_aggregate(processed_results: List[int]):
+
+    num_exception = 0
+    num_correct = 0
+
+    for i in range(len(processed_results)):
+        if processed_results[i] == -1:
+            num_exception += 1
+        elif processed_results[i] == 1:
+            num_correct += 1
+
+    num_total = len(processed_results)
+    accuracy = round(1.0 * num_correct / num_total, 2)
+    error_rate = round(1.0 * num_exception / num_total, 2)
+
+    log_metric(key="accuracy", value=accuracy)
+    log_metric(key="error_rate", value=error_rate)
+
+    return {
+        "num_total": num_total,
+        "num_correct": num_correct,
+        "num_exception": num_exception,
+        "accuracy": accuracy,
+        "error_rate": error_rate
+    }
+
+
+if __name__ == "__main__":
+    numbers = [1, 1, 1, 1, 0, -1, -1]
+    accuracy = accuracy_aggregate(numbers)
+    print("The accuracy is", accuracy)

examples/flows/evaluation/eval-accuracy-maths-to-code/flow.dag.yaml

@@ -0,0 +1,29 @@
+$schema: https://azuremlschemas.azureedge.net/promptflow/latest/Flow.schema.json
+inputs:
+  groundtruth:
+    type: string
+    default: "1"
+  prediction:
+    type: string
+    default: "2"
+outputs:
+  score:
+    type: string
+    reference: ${line_process.output}
+nodes:
+- name: line_process
+  type: python
+  source:
+    type: code
+    path: line_process.py
+  inputs:
+    groundtruth: ${inputs.groundtruth}
+    prediction: ${inputs.prediction}
+- name: aggregate
+  type: python
+  source:
+    type: code
+    path: aggregate.py
+  inputs:
+    processed_results: ${line_process.output}
+  aggregation: true

examples/flows/evaluation/eval-accuracy-maths-to-code/line_process.py

@@ -0,0 +1,40 @@
+from promptflow.core import tool
+
+
+@tool
+def line_process(groundtruth: str, prediction: str) -> int:
+
+    processed_result = 0
+
+    if prediction == "JSONDecodeError" or prediction.startswith("Unknown Error:"):
+        processed_result = -1
+        return processed_result
+
+    try:
+        groundtruth = float(groundtruth)
+        prediction = float(prediction)
+    except ValueError:
+        processed_result = -1
+        return processed_result
+
+    if round(prediction, 2) == round(groundtruth, 2):
+        processed_result = 1
+
+    return processed_result
+
+
+if __name__ == "__main__":
+    processed_result = line_process("1.0", "1")
+    print("The processed result is", processed_result)
+
+    processed_result = line_process("3.14", "3.1415926")
+    print("The processed result is", processed_result)
+
+    processed_result = line_process("2.1", "2.0")
+    print("The processed result is", processed_result)
+
+    processed_result = line_process("1.0", "JSONDecodeError")
+    print("The processed result is", processed_result)
+
+    processed_result = line_process("1.0", "No module named 'numpy'")
+    print("The processed result is", processed_result)

examples/flows/evaluation/eval-accuracy-maths-to-code/test_data.jsonl

@@ -0,0 +1,20 @@
+{"question": "What is the sum of 5 and 3?", "groundtruth": "8", "answer": "8"}
+{"question": "Subtract 7 from 10.", "groundtruth": "3", "answer": "3"}
+{"question": "Multiply 6 by 4.", "groundtruth": "24", "answer": "24"}
+{"question": "Divide 20 by 5.", "groundtruth": "4", "answer": "4"}
+{"question": "What is the square of 7?", "groundtruth": "49", "answer": "49"}
+{"question": "What is the square root of 81?", "groundtruth": "9", "answer": "9"}
+{"question": "If a rectangle has a length of 10 and width of 5, what is the area?", "groundtruth": "50", "answer": "50"}
+{"question": "A circle has a radius of 7, what is the area? (Use 3.14 for pi)", "groundtruth": "153.86", "answer": "153.871"}
+{"question": "Solve for x in the equation 2x + 3 = 9.", "groundtruth": "3", "answer": "3"}
+{"question": "What is the value of x if 5x = 25?", "groundtruth": "5", "answer": "5"}
+{"question": "A car travels 200 miles in 4 hours. What is the average speed of the car?", "groundtruth": "50", "answer": "50"}
+{"question": "A car travels at a speed of 60 mph. How long will it take to travel 180 miles?", "groundtruth": "3", "answer": "3"}
+{"question": "If a car travels at a speed of 40 mph for 2 hours, how far will it travel?","groundtruth": "80", "answer": "80"}
+{"question":"A rectangle has length = 10 cm and width = 5 cm. What is its area?", "groundtruth":"50", "answer": "50"}
+{"question":"A circle has radius = 7 cm. What is its circumference? (Use pi =3.14)", "groundtruth":"43.96", "answer": "43.959"}
+{"question":"A triangle has base =10 cm and height =5 cm. What is its area?", "groundtruth":"25", "answer": "25"}
+{"question":"What is the slope of the line that passes through (2,3) and (4,7)?", "groundtruth":"2", "answer": "2"}
+{"question":"The distance between A and B is 2000km, A is moving towards B with speed 80km/hour, meanwhile B is moving towards A with speed 120km/hour, how many hours later A and B can meet?", "groundtruth":"10", "answer": "10"}
+{"question":"The lengths of the two perpendicular sides of a right triangle are 6cm and 8cm. What is the length of the hypotenuse?", "groundtruth": "10", "answer": "10"}
+{"question":"A is running with average speed 10km/hour, A already run half hour. B start to chase A along the same route with average speed 15km/hour, how many hours B will take to meet A?", "groundtruth":"1", "answer": "2"}