[Playground] Better visualize how much text is returned by adding skeleton text to represent max tokens variable

Hello OpenAI community members!

It is said that “one token is roughly 4 characters for normal English text,” [Source - Docs] I still find it hard to actually visualize the amount of text in a completion. I’d like to provide feedback that could help make it easier for users to better understand how much text could be returned in a completion when adjusting the Maximum Length variable.

I could multiply 40 tokens by 4 to estimate that the completion will be 160 characters or less, but I also find that hard to visualize, since I’m used to seeing how much space a block of text takes up on a page rather than trying to visualize how much space is used up by a block of characters.

To help users who also see that as a struggle, I think we can more accurately represent how much text could be returned by having skeleton placeholder text that represents the number of tokens a user has selected.

This should help users better visualize the amount of text they might see in a completion. Here’s an example below to understand what I mean:

Let me know if you all would find this useful, or if you have any concerns regarding this potential feature.

1 Like

Welcome back @DutytoDevelop

Hope you’ve been doing well.

It’s a very interesting idea, given how we can’t tell exactly how many characters tokens would translate to.

However I think that it would only hold for English, given how tokenizer is based on English.

1 Like