Hey guys, I am using Genkit to embed
Hey guys, I am using Genkit to embed text, then save the vector collection in Firestore. I have embedded a fair amount of data: 100,000 vectors in 100,000 docs.
My concern is, when searching it takes too much to return a response, around 20 secs, this of course it is not acceptable for clients specially for search.
this is how my retriever looks
using
any idea why retrieving takes too long? is it cause of the amount of items, or because I am using Firestore as the sort of Vector DB does Genkit indeed supports
My concern is, when searching it takes too much to return a response, around 20 secs, this of course it is not acceptable for clients specially for search.
this is how my retriever looks
defineFirestoreRetriever(this.ai, {
name: HS_RETRIEVER_NAME,
firestore: this.db,
collection: PRODUCTS_100K_DATASET,
contentField: CONTENT_FIELD,
vectorField: VECTOR_FIELD,
embedder: textEmbeddingGecko003,
distanceMeasure: "DOT_PRODUCT",
metadataFields: ["id", "title"],
}); using
genkit 1.4.0any idea why retrieving takes too long? is it cause of the amount of items, or because I am using Firestore as the sort of Vector DB does Genkit indeed supports
