| import csv |
| import json |
|
|
| input_file = 'dataset.csv' |
| output_file = 'dataset_alpaca.jsonl' |
|
|
| with open(input_file, 'r', encoding='utf-8') as csvfile, open(output_file, 'w', encoding='utf-8') as jsonlfile: |
| reader = csv.DictReader(csvfile) |
| for row in reader: |
| emotion_in = row.get('emmotion_in', '').strip() |
| user_input = row.get('input', '').strip() |
| context = row.get('context', '').strip() |
| output = row.get('output', '').strip() |
|
|
| instruction = f"The user feels an emotion of '{emotion_in}' and said : \"{user_input}\"" |
| input_text = (f"\nContext : {context}" if context else "") |
| |
| prompt = { |
| "instruction": instruction, |
| "input": input_text, |
| "output": output |
| } |
| jsonlfile.write(json.dumps(prompt, ensure_ascii=False) + '\n') |
|
|
