I am learning the basics of vector search in Vespa and went through https://github.com/vespa-engine/sample-apps/blob/master/msmarco-ranking/passage-ranking-README.md . I studied the four-part blog too. Now it would be super interesting to replace the sample dataset colmini-passage-feed-sample.jsonl with a similar sample dataset of my own. Do you have a script, or anything more concrete than the blog part 3?
Asked
Active
Viewed 104 times
2
Roope K
- 101
- 5
1 Answers
3
This worked.
{"put": "id:msmarco:passage::1001", "fields": {"id": 0, "text": "Vespa is great!"}}
We don't actually have to calculate the tensors outside of Vespa.
From Jo Kristian:
If you bring your own text and without embedding or multi/embedding, the sample app produces it during document processing
Roope K
- 101
- 5