LLM Evaluation Report

Model
Date
Total Response Time (s)
Tests Passed
Mean CodeBLEU (0-1)
Mean Usefulness Score (0-4)
Mean Functional Correctness Score (0-4)

claude-opus-4-20250514

2025-05-27

682.341

45

0.373498

3.68902

3.71951

claude-sonnet-4-20250514

2025-05-27

685.546

112

0.317174

3.7378

3.65854

claude-3-7-sonnet-20250219

2025-05-27

746.497

108

0.319258

3.65244

3.65244

claude-3-5-sonnet-20241022

2025-05-27

445.549

114

0.332094

3.65244

3.72561

gpt-4.1

2025-05-27

340.45

114

0.345565

3.71951

3.79878

o4-mini

2025-05-27

1380.26

128

0.322408

3.70122

3.7439

o3

2025-05-27

1592.45

141

0.314449

3.71341

3.85366

gpt-4o

2025-05-27

254.478

123

0.305002

3.70732

3.7378

gemini_gemini-2.0-flash

2025-05-27

428.324

102

0.304022

3.65244

3.60976

gemini_gemini-2.5-pro-preview-05-06

2025-05-27

1317.42

71

0.319577

2.45732

2.67683

gemini_gemini-2.5-flash-preview-05-20

2025-05-27

1042.03

108

0.32728

3.39024

3.46341

Total Response Time (s): The total time taken by the model to generate all the outputs.

Tests passed: The number of unit tests that the model has passed during evaluation, out of a total of 164 tests.

Mean CodeBLEU: Average CodeBLEU score, a metric for evaluating code generation quality based on both syntactic and semantic correctness.

Mean Usefulness Score: Average rating of the model's output usefulness as rated by a LLM model.

  • 0: Snippet is not at all helpful, it is irrelevant to the problem.

  • 1: Snippet is slightly helpful, it contains information relevant to the problem, but it is easier to write the solution from scratch.

  • 2: Snippet is somewhat helpful, it requires significant changes (compared to the size of the snippet), but is still useful.

  • 3: Snippet is helpful, but needs to be slightly changed to solve the problem.

  • 4: Snippet is very helpful, it solves the problem.

Mean Functional Correctness Score: Average score of the functional correctness of the model's outputs, assessing how well the outputs meet the functional requirements, rated by a LLM model.

  • 0 (failing all possible tests): The code snippet is totally incorrect and meaningless.

  • 4 (passing all possible tests): The code snippet is totally correct and can handle all cases.

Last updated