Why does Azure Open AI embedding not allow to an array of text as input

API
1

Hi All,
Does anyone know why Azure Open AI do not support this while Open AI does. I cannot say

create_embed([“text1”,“text2”] with Azure Open AI

2

Can you show the code and the errors you receive? (the example you posted is missing a closing “)”)

Also what libraries are you using.

3

Yes, that is a typo. The error that I get is

Embedding process failed with too many inputs. the max number of inputs is 1. We hope to increase the number of inputs per request soon.

4

Ok, well for now it seems you are limited to one with the OpenAI call. What advantage do you get from adding more than one?

5

What advantage do you get from adding more than one?

Batching multiple texts in a single request is useful when embedding large amounts of text. In addition to increasing speed of embedding, reducing the number of requests reduces the chance of exceeding rate limits.

1 Like
6

I have 1100 documents which I chunk, average chunk number is around 600. I had to do one by one call to the API. If I can send a bunch together, I think it will reduce the network hops

7

OK, so the embedding models are so fast that network overheads are a factor, ok cool, thanks for the info.