Generate a list of experiences for testing.
(task_num: int, repeat_times: int = 1, step_num: int = 1)
| 10 | |
| 11 | |
| 12 | def get_experiences(task_num: int, repeat_times: int = 1, step_num: int = 1) -> List[Experience]: |
| 13 | """Generate a list of experiences for testing.""" |
| 14 | return [ |
| 15 | Experience( |
| 16 | eid=EID(task=i, run=j, step=k), |
| 17 | tokens=torch.zeros((5,)), |
| 18 | prompt_length=4, |
| 19 | reward=j, |
| 20 | logprobs=torch.tensor([0.1]), |
| 21 | info={ |
| 22 | "llm_quality_score": i, |
| 23 | "llm_difficulty_score": k, |
| 24 | }, |
| 25 | ) |
| 26 | for i in range(task_num) |
| 27 | for j in range(repeat_times) |
| 28 | for k in range(step_num) |
| 29 | ] |
| 30 | |
| 31 | |
| 32 | class TestRewardShapingMapper(unittest.IsolatedAsyncioTestCase): |
no test coverage detected