1. DB Benchmark:
- Now you can improve the quality of vectors in the database using the new DB Benchmark page. Simply enter your query, select the number of vectors (from 1 to 10), and start the process. The result will be a list of the nearest vectors to your query, which can be edited and updated in the knowledge database.
- Enrichment Feature: Generates 10 queries for each given vector, which can be edited and updated in the database. This helps to expand the knowledge base more quickly and efficiently.
2. Model Testing:
- On the new Model Testing page, you can test the quality of our model not only based on the vector database but also based on input prompts and chat with GPT. You can get detailed information about the generated response, including the vectors used, their distance, the prompt, and the number of tokens spent on generating the response.
- Finally, if you like the generated answer, you can save it as a new vector.
3. New Messenger for "Custom chat":
- Now you can easily and quickly integrate our new messenger for "Custom chat" into your website to provide convenient communication with your users.
This is what the chat will look like on your site:
4. Ability to Choose Knowledge Base for Each Messenger Separately:
- Now you can select a knowledge base separately for each messenger, allowing you to more precisely tailor interactions with users.
5. Update to Training Page:
- The Training page has been updated to improve the user experience in training and getting the model up and running effectively. It now includes the ability to view all vectors, with a resource label, and individually view the vectors of each resource.
