| 52 | |
| 53 | |
| 54 | def preprocess_dataset( |
| 55 | dataset: Dataset, |
| 56 | template: Template, |
| 57 | reverse_template: Template, |
| 58 | option_a, |
| 59 | option_b, |
| 60 | initial_instruct: str, |
| 61 | reversed_initial_instruct: str, |
| 62 | ): |
| 63 | train_data = list() |
| 64 | test_data = list() |
| 65 | for item in dataset['train']: |
| 66 | train_data.append([ |
| 67 | template.substitute( |
| 68 | passage=item['prompt'], |
| 69 | option_a=option_a, |
| 70 | option_b=option_b, |
| 71 | initial_instruct_passage=initial_instruct |
| 72 | ), |
| 73 | template.substitute( |
| 74 | passage=item['prompt'], |
| 75 | option_a=option_b, |
| 76 | option_b=option_a, |
| 77 | initial_instruct_passage=reversed_initial_instruct |
| 78 | ), |
| 79 | reverse_template.substitute( |
| 80 | passage=item['prompt'], |
| 81 | option_a=option_a, |
| 82 | option_b=option_b, |
| 83 | initial_instruct_passage=initial_instruct |
| 84 | ), |
| 85 | reverse_template.substitute( |
| 86 | passage=item['prompt'], |
| 87 | option_a=option_b, |
| 88 | option_b=option_a, |
| 89 | initial_instruct_passage=reversed_initial_instruct |
| 90 | ), |
| 91 | item['helpful'] |
| 92 | ]) |
| 93 | for item in dataset['test']: |
| 94 | test_data.append([ |
| 95 | template.substitute( |
| 96 | passage=item['prompt'], |
| 97 | option_a=option_a, |
| 98 | option_b=option_b, |
| 99 | initial_instruct_passage=initial_instruct |
| 100 | ), |
| 101 | template.substitute( |
| 102 | passage=item['prompt'], |
| 103 | option_a=option_b, |
| 104 | option_b=option_a, |
| 105 | initial_instruct_passage=reversed_initial_instruct |
| 106 | ), |
| 107 | reverse_template.substitute( |
| 108 | passage=item['prompt'], |
| 109 | option_a=option_a, |
| 110 | option_b=option_b, |
| 111 | initial_instruct_passage=initial_instruct |