I have a dataset that contains columns in the form of [‘title’, ‘subtitle’, ‘content’]. When creating embeddings, I only send content to embed, do you think this approach is correct? When adding data to the vector database, I use the title and subtitle columns to indexing.
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Indexing embeddings when storing them in the vector database | 2 | 1902 | March 25, 2023 | |
| Preparing the dataset for embeddings | 10 | 6140 | December 17, 2023 | |
| How to use text embedding ada 002 with a dataset that contains four hundred rows | 2 | 1449 | June 24, 2023 | |
| I want to known about Text Search using text-embedding-ada-002? | 3 | 1096 | December 20, 2023 | |
| Help needed with embedding recommendations and multiple databases | 2 | 727 | January 21, 2023 |