You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Within the current version, we leverage the users selected large language model to extract package / module / library names from code snippets. While this was a good approach for our first version, it maybe worth considering other options, as a level of overhead comes from the large language models processing times. When combined with embeddings and vectorization, it appears it has a noticeable impact on performance within the users path / experience.
In the previous prototype , tree sitter was used, but the library is lacking support for wide library coverege, so a different approach maybe required.
This Story is to allow research over alternative approaches, proof of concept to demonstrate a notable performance improvement and then implementation
The text was updated successfully, but these errors were encountered:
Within the current version, we leverage the users selected large language model to extract package / module / library names from code snippets. While this was a good approach for our first version, it maybe worth considering other options, as a level of overhead comes from the large language models processing times. When combined with embeddings and vectorization, it appears it has a noticeable impact on performance within the users path / experience.
In the previous prototype , tree sitter was used, but the library is lacking support for wide library coverege, so a different approach maybe required.
This Story is to allow research over alternative approaches, proof of concept to demonstrate a notable performance improvement and then implementation
The text was updated successfully, but these errors were encountered: