We have a view on these; we can DM to exchange specifics on your index to help you mitigate if you c

JJerome We have a view on these; we can DM to exchange specifics on your index to help y...

D

divby0•11/28/23, 11:39 AM

that would be awesome. I'll DM you the details in a few minutes

E

Esked•11/29/23, 5:58 PM

Hello! I'm having some issues with Vectorize's query function.
When I run

await env.<INDEX>.query(vector, { topK: 1})

await env.<INDEX>.query(vector, { topK: 1})

await env.<INDEX>.query(vector, { topK: 1})

await env.<INDEX>.query(vector, { topK: 1}), in some cases it simply fails and returns the following error message:

TypeError: Cannot read properties of undefined (reading '0')

TypeError: Cannot read properties of undefined (reading '0')

TypeError: Cannot read properties of undefined (reading '0')

TypeError: Cannot read properties of undefined (reading '0')
It's completely unpredictable, sometimes it works perfectly after a couple of calls, sometimes it fails on the very first one. The

reading '0'

reading '0'

reading '0'

reading '0' isn't constant either, sometimes the number is different, and different userQueries result in different numbers.
Here is the code for the function

export async function getEmbedResponse(env: Environment, userQuery: string, data: string[]): Promise<{ score: number, result: string }> {
  const ai = new Ai(env.AI)

  const queryVector: EmbeddingResponse = await ai.run('@cf/baai/bge-base-en-v1.5', {
    text: [userQuery],
  })

  let matches = await env.<INDEX>.query(queryVector.data[0], { topK: 1 })

  const embedAiResponse = { score: matches.matches[0].score, result: data[parseInt(matches.matches[0].vectorId)] }

  return embedAiResponse
}

export async function getEmbedResponse(env: Environment, userQuery: string, data: string[]): Promise<{ score: number, result: string }> {
  const ai = new Ai(env.AI)

  const queryVector: EmbeddingResponse = await ai.run('@cf/baai/bge-base-en-v1.5', {
    text: [userQuery],
  })

  let matches = await env.<INDEX>.query(queryVector.data[0], { topK: 1 })

  const embedAiResponse = { score: matches.matches[0].score, result: data[parseInt(matches.matches[0].vectorId)] }

  return embedAiResponse
}

export async function getEmbedResponse(env: Environment, userQuery: string, data: string[]): Promise<{ score: number, result: string }> {
  const ai = new Ai(env.AI)

  const queryVector: EmbeddingResponse = await ai.run('@cf/baai/bge-base-en-v1.5', {
    text: [userQuery],
  })

  let matches = await env.<INDEX>.query(queryVector.data[0], { topK: 1 })

  const embedAiResponse = { score: matches.matches[0].score, result: data[parseInt(matches.matches[0].vectorId)] }

  return embedAiResponse
}

export async function getEmbedResponse(env: Environment, userQuery: string, data: string[]): Promise<{ score: number, result: string }> {
  const ai = new Ai(env.AI)

  const queryVector: EmbeddingResponse = await ai.run('@cf/baai/bge-base-en-v1.5', {
    text: [userQuery],
  })

  let matches = await env.<INDEX>.query(queryVector.data[0], { topK: 1 })

  const embedAiResponse = { score: matches.matches[0].score, result: data[parseInt(matches.matches[0].vectorId)] }

  return embedAiResponse
}

EmbeddingResponse is an interface based on the examples on the website

interface EmbeddingResponse {
    shape: number[];
    data: number[][];
}

interface EmbeddingResponse {
    shape: number[];
    data: number[][];
}

interface EmbeddingResponse {
    shape: number[];
    data: number[][];
}

interface EmbeddingResponse {
    shape: number[];
    data: number[][];
}

E

Esked•11/29/23, 5:58 PM

I run the vector upserting before calling this function

E

Esked•11/29/23, 8:46 PM

I can see the error message after the query call too

E

Esked•11/29/23, 8:49 PM

E

Esked•11/29/23, 8:49 PM

(edited image to cover information, oops

)

E

Esked•11/29/23, 9:06 PM

Ok I've tested a bit more and it seems like the problem is on my end.
The

data

data

data

data array is sometimes

undefined

undefined

, not sure why but that should be my problem

E

Esked•11/29/23, 9:18 PM

Yep, the problem is definitely on my end, sorry

D

divby0•11/30/23, 8:46 AM

@Jerome have you gotten any info on what the plans for vectorize bindings inside pages functions are? Asking here instead of DM because others are probably looking ofr that info as well

Ddivby0 @Jerome have you gotten any info on what the plans for vectorize bindings inside...

M

Matt Silverlock•11/30/23, 3:33 PM

Soon! Ideally before the next holiday period. Prioritizing some stability/scalability work and metadata filtering first, but should be a very fast follow.

C

chand1012•11/30/23, 7:57 PM

Is there a way when querying vectorize to filter by ids as well as the vector themselves? You mentioned metadata filtering, I guess that's pretty close to what I'm looking for. I formatted my IDs in such a way where they contain both an index and the ID of them in my D1 database.

MMatt Silverlock Soon! Ideally before the next holiday period. Prioritizing some stability/scalab...

D

divby0•11/30/23, 9:34 PM

Aweomse!

Cchand1012 Is there a way when querying vectorize to filter by ids as well as the vector th...

H

huw•12/1/23, 3:24 AM

Yeah, this would be supported by metadata filtering or even namespaces. If you only ever want to filter on one of your ‘indices’ at a time, you could totally use namespaces today; for anything more complex you should wait for metadata filtering

Ccfnathan Noted as a feature request! Just so we're on the same page: would metadata filte...

C

chand1012•12/1/23, 3:56 PM

Metadata filtering would be fine, do you have an ETA on when this feature will be ready?

C

chand1012•12/1/23, 4:16 PM

My use case for either feature is I'd like to have one vector index for all the items in a database table, when some of the content from the table may be from both user A and user B, and I only want to query for user A as they're the one making the request.

S

steyblind•12/1/23, 5:08 PM

Not sure if I missed something here:

1. metadata "included in results" was released
2. on an existing worker, I updated the compatibility date
3. on an existing vectorize db there was no change
4. I updated my worker to use the

includeMetadata

includeMetadata

includeMetadata

includeMetadata param

and... no metadata returned?

Exception Thrown @ 12/1/2023, 12:06:34 PM
  (log) {
  count: 2,
  matches: [
    { vectorId: 'hzz2v', score: 0.804275747, vector: [Object] },
    { vectorId: 'aml6o', score: 0.726804477, vector: [Object] }
  ]
}

Exception Thrown @ 12/1/2023, 12:06:34 PM
  (log) {
  count: 2,
  matches: [
    { vectorId: 'hzz2v', score: 0.804275747, vector: [Object] },
    { vectorId: 'aml6o', score: 0.726804477, vector: [Object] }
  ]
}

Exception Thrown @ 12/1/2023, 12:06:34 PM
  (log) {
  count: 2,
  matches: [
    { vectorId: 'hzz2v', score: 0.804275747, vector: [Object] },
    { vectorId: 'aml6o', score: 0.726804477, vector: [Object] }
  ]
}

Exception Thrown @ 12/1/2023, 12:06:34 PM
  (log) {
  count: 2,
  matches: [
    { vectorId: 'hzz2v', score: 0.804275747, vector: [Object] },
    { vectorId: 'aml6o', score: 0.726804477, vector: [Object] }
  ]
}

Cchand1012 Metadata filtering would be fine, do you have an ETA on when this feature will b...

V

Vy•12/1/23, 5:08 PM

we're hoping it'll be out in December, so stay tuned!

Ssteyblind Not sure if I missed something here: 1. metadata "included in results" was rele...

V

Vy•12/1/23, 5:09 PM

metadata filtering is not available yet, we're hoping to make it available for new indexes in Dec, so stay tuend

S

steyblind•12/1/23, 5:09 PM

sorry, I mispoke

S

steyblind•12/1/23, 5:09 PM

I meant metadata in results

S

steyblind•12/1/23, 5:09 PM

per vectorize @vaiton13

S

steyblind•12/1/23, 5:15 PM

let contextResult = await env.DEMO_VECTORDB.query(queryVector, { topK: 5, returnMetadata: true })

let contextResult = await env.DEMO_VECTORDB.query(queryVector, { topK: 5, returnMetadata: true })

let contextResult = await env.DEMO_VECTORDB.query(queryVector, { topK: 5, returnMetadata: true })

let contextResult = await env.DEMO_VECTORDB.query(queryVector, { topK: 5, returnMetadata: true })

S

steyblind•12/1/23, 5:16 PM

(the exception is thrown when I try to access the metadata prop)

S

steyblind•12/1/23, 7:27 PM

thank you! good to know it's not me

S

steyblind•12/1/23, 7:27 PM

S

steyblind•12/1/23, 7:27 PM

godspeed

S

steyblind•12/1/23, 7:27 PM

I'll go back to metadata in KV

D

dave•12/2/23, 8:44 PM

https://github.com/langchain-ai/langchain/issues/14179

GitHub

Cloudflare Vectorize Support · Issue #14179 · langchain-ai/langchain

Feature request Looks like it should be doable to have Cloudflare Vectorize support, as there's a HTTP API for Vectorize. https://developers.cloudflare.com/api/operations/vectorize-list-vectori...

D

dave•12/2/23, 8:44 PM

if anybody official wants to chime in on the ticket to confirm if my assumptions are wrong/right

Ddave https://github.com/langchain-ai/langchain/issues/14179

M

Matt Silverlock•12/4/23, 2:28 PM

See https://js.langchain.com/docs/integrations/text_embedding/cloudflare_ai

Cloudflare Workers AI | 🦜️🔗 Langchain

If you're deploying your project in a Cloudflare worker, you can use Cloudflare's built-in Workers AI embeddings with LangChain.js.

M

Matt Silverlock•12/4/23, 2:28 PM

import { CloudflareVectorizeStore } from "langchain/vectorstores/cloudflare_vectorize";

import { CloudflareVectorizeStore } from "langchain/vectorstores/cloudflare_vectorize";

import { CloudflareVectorizeStore } from "langchain/vectorstores/cloudflare_vectorize";

import { CloudflareVectorizeStore } from "langchain/vectorstores/cloudflare_vectorize";

M

Matt Silverlock•12/4/23, 2:28 PM

and https://js.langchain.com/docs/integrations/vectorstores/cloudflare_vectorize

Cloudflare Vectorize | 🦜️🔗 Langchain

If you're deploying your project in a Cloudflare worker, you can use Cloudflare Vectorize with LangChain.js.

MMatt Silverlock See https://js.langchain.com/docs/integrations/text_embedding/cloudflare_ai

D

dave•12/4/23, 2:28 PM

Oh I know, I just want it to use Vectorize in Python too!

M

Matt Silverlock•12/4/23, 2:29 PM

Ah, sorry, misread. You should contribute it

U

umair mehmood•12/5/23, 9:47 AM

I am getting this when creating vector index from API
Payload

{
  "config": {
    "preset": "@cf/baai/bge-small-en-v1.5",
    "dimensions": 768,
    "metric": "cosine"

  },
  "description": "This is my example index.",
  "name": "example-index"
}

{
  "config": {
    "preset": "@cf/baai/bge-small-en-v1.5",
    "dimensions": 768,
    "metric": "cosine"

  },
  "description": "This is my example index.",
  "name": "example-index"
}

{
  "config": {
    "preset": "@cf/baai/bge-small-en-v1.5",
    "dimensions": 768,
    "metric": "cosine"

  },
  "description": "This is my example index.",
  "name": "example-index"
}

{
  "config": {
    "preset": "@cf/baai/bge-small-en-v1.5",
    "dimensions": 768,
    "metric": "cosine"

  },
  "description": "This is my example index.",
  "name": "example-index"
}

response

{
    "result": null,
    "success": false,
    "errors": [
        {
            "code": 1005,
            "message": "vectorize.not_entitled"
        }
    ],
    "messages": []
}

{
    "result": null,
    "success": false,
    "errors": [
        {
            "code": 1005,
            "message": "vectorize.not_entitled"
        }
    ],
    "messages": []
}

{
    "result": null,
    "success": false,
    "errors": [
        {
            "code": 1005,
            "message": "vectorize.not_entitled"
        }
    ],
    "messages": []
}

{
    "result": null,
    "success": false,
    "errors": [
        {
            "code": 1005,
            "message": "vectorize.not_entitled"
        }
    ],
    "messages": []
}

K

kian•12/5/23, 9:50 AM

Does that account have a Workers Paid plan?

U

umair mehmood•12/5/23, 9:56 AM

No

U

umair mehmood•12/5/23, 9:57 AM

I see now its avilable on paid plan only.

U

umair mehmood•12/5/23, 9:58 AM

btw from where I can get the

account_identifier

account_identifier

account_identifier

account_identifier?

K

kian•12/5/23, 10:15 AM

https://developers.cloudflare.com/fundamentals/setup/find-account-and-zone-ids/

Find zone and account IDs · Cloudflare Fundamentals docs

Once you set up a new account and add your domain to Cloudflare, you may need access to your zone and account IDs for API operations.

K

kian•12/5/23, 10:16 AM

It'll be in the URL when you're on the dashboard, but otherwise you can find it on the sidebar as the docs say.

Ccfnathan There is currently a bug with this: fix is here - https://github.com/cloudflare/...

S

steyblind•12/5/23, 2:14 PM

this is fixed, but awaiting release, is that correct?

V

Vy•12/6/23, 5:53 PM

Hi folks, it's a fun day because metadata filtering is now available with Vectorize! Check out the docs here, https://developers.cloudflare.com/vectorize/learning/metadata-filtering/. Happy building and please share any feedback or issues!

Metadata filtering · Vectorize

In addition to providing an input vector to your query, you can also filter by vector metadata associated with every vector. Query results only …

VVy Hi folks, it's a fun day because metadata filtering is now available with Vector...

S

scotto•12/7/23, 12:12 AM

awesome! namespace filtering is before than metadata filtering?

Sscotto awesome! namespace filtering is before than metadata filtering?

J

JeromeOP•12/7/23, 9:33 AM

yes, but it shouldn't matter regarding the set of vectors in the response, as the intersection of both match sets is what's taken into consideration when results are evaluated

We have a view on these; we can DM to exchange specifics on your index to help you mitigate if you c

Similar Threads