I’m doing embeddings using ADA engine with openai.Embedding.create. In the billing history i noticed more requests then those really performed. How is it possible?
I’ve a pandas dataframe with 3700 records, at about 11.00 i’ve started request embedding for each record but this is what i read from my user dashboard
Why so many request for 3700 records?
In total you have roughly 25k tokens used, 25k tokens is very roughly 1/4 of the size of the data you embeded, meaning you embedded about 100 Kilobytes of data +/- 50 Kilobytes depending on content.
So if you look at your record sizes in total, they should be around 50-150Kb in size.
Hope this helps.
It was my fault!! 
I’ve created a loop in my application while requesting embeddings!!
Thank’s anyway 
