1 Answer
- Newest
- Most votes
- Most comments
0
Hi, Are you trying to do fine-tuning or continuous pre-training. They have different expected format. The format required for continued pre-training is : {"input": "<input text>"} {"input": "<input text>"} {"input": "<input text>"}
The format for fine-tuning is: {"prompt": "<prompt1>", "completion": "<expected generated text>"} {"prompt": "<prompt2>", "completion": "<expected generated text>"} {"prompt": "<prompt3>", "completion": "<expected generated text>"}
For further guidance refer here: [https://docs.aws.amazon.com/bedrock/latest/userguide/model-customization-prepare.html]
answered 9 days ago
Relevant content
- asked 2 months ago
- AWS OFFICIALUpdated 2 months ago
- AWS OFFICIALUpdated 4 months ago
- AWS OFFICIALUpdated a month ago
- AWS OFFICIALUpdated 18 days ago
Would be very useful to provide you command and file name. You can redact part of the bucket name to maintain security.