dify icon indicating copy to clipboard operation
dify copied to clipboard

The number of segments generated(QA format) is 0

Open itltf512116 opened this issue 1 year ago • 3 comments

Self Checks

Dify version

0.3.34

Cloud or Self Hosted

Self Hosted (Docker)

Steps to reproduce

After generating a document in template.csv format, and selecting the QA format when submitting the segments, the number of segments generated is 0. Is this because of formatting issues? What should be the correct format?

✔️ Expected Behavior

Correctly parses document content

❌ Actual Behavior

The number of segments generated is 0

itltf512116 avatar Dec 28 '23 01:12 itltf512116

Have you finished the whole process, can to make a screenshot of the datasets in this document?

crazywoola avatar Dec 29 '23 01:12 crazywoola

@crazywoola Thanks for your reply,the screenshot files is: 14035 2046 template.csv

Have you finished the whole process, can to make a screenshot of the datasets in this document?

itltf512116 avatar Dec 29 '23 02:12 itltf512116

For example template.txt

Q: question 1 A: answer 1 Q: question 2 A: answer 2

Since the document is already in Q&A format, you can turn off the Q&A switch. And use the format above. Also, you can use custom splitter\n.

crazywoola avatar Dec 30 '23 02:12 crazywoola

For example template.txt

Q: question 1 A: answer 1 Q: question 2 A: answer 2

Since the document is already in Q&A format, you can turn off the Q&A switch. And use the format above. Also, you can use custom splitter\n.

Hi @crazywoola ,I've reorganized the contents of the file according to your formatting, knowledge.txt The screenshot is as follows: image

The number of segments is now correct, but the format of the segment content doesn't seem to be in QA format: image

itltf512116 avatar Jan 02 '24 01:01 itltf512116

QA format is to generate new questions based on your document segments.

JohnJyong avatar Jan 05 '24 10:01 JohnJyong