From 9b656762c45bac86a4c8c48c1dbd0e9f82cf39b4 Mon Sep 17 00:00:00 2001 From: Claas Augner <495429+caugner@users.noreply.github.com> Date: Fri, 5 Apr 2024 12:44:32 +0200 Subject: [PATCH] feat(ai-help): switch to text-embedding-3-small model (#459) --- src/ai/constants.rs | 2 +- src/ai/embeddings.rs | 6 +++--- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/src/ai/constants.rs b/src/ai/constants.rs index f268eaec..f5154ff0 100644 --- a/src/ai/constants.rs +++ b/src/ai/constants.rs @@ -51,7 +51,7 @@ pub const AI_HELP_GPT4_FULL_DOC_NEW_PROMPT: AIHelpConfig = AIHelpConfig { }; pub const MODEL: &str = "gpt-3.5-turbo"; -pub const EMBEDDING_MODEL: &str = "text-embedding-ada-002"; +pub const EMBEDDING_MODEL: &str = "text-embedding-3-small"; pub const AI_HELP_SYSTEM_MESSAGE: &str = "You are a very enthusiastic MDN AI who loves \ to help people! Given the following information from MDN, answer the user's question \ diff --git a/src/ai/embeddings.rs b/src/ai/embeddings.rs index d6d98db4..b3cd0743 100644 --- a/src/ai/embeddings.rs +++ b/src/ai/embeddings.rs @@ -49,13 +49,13 @@ const MACRO_DOCS_QUERY: &str = "SELECT doc.title, parent.title_short AS title_parent, doc.markdown AS content, - doc.embedding <=> $1 AS similarity + doc.embedding_next <=> $1 AS similarity FROM mdn_doc_macro doc LEFT JOIN mdn_doc_macro parent ON parent.mdn_url = SUBSTRING(doc.mdn_url, 1, LENGTH(doc.mdn_url) - STRPOS(REVERSE(doc.mdn_url), '/')) WHERE LENGTH(doc.markdown) >= $4 - AND (doc.embedding <=> $1) < $2 + AND (doc.embedding_next <=> $1) < $2 AND doc.mdn_url NOT LIKE '/en-US/docs/MDN%' -ORDER BY doc.embedding <=> $1 +ORDER BY doc.embedding_next <=> $1 LIMIT $3;"; #[derive(sqlx::FromRow, Debug)]