Training foundation models · Question 10 of 12
Which data format does fine-tuning in Amazon Bedrock typically expect?
1 answer is correct. Medium
Show answer
Answer
- D JSONL files with input-output pairs in Amazon S3
Explanation
For fine-tuning, training data is usually provided as JSON Lines in S3, with one example per line consisting of input (prompt) and desired output. For continued pre-training unlabelled texts are sufficient. The exact format depends on the model.