Show HN: I made a handy emoji picker on macOS with GPT-3(getmumu.com)
getmumu.com
Show HN: I made a handy emoji picker on macOS with GPT-3
https://getmumu.com
7 コメント
Hi! The founder here.
I use GPT-3 to generate synonyms for all of the emojis.
And yes, it’s simple to have a list of synonyms. But is it easy? Calculate the time anyone takes to write synonyms for a particular emoji, and multiply it by the number of emojis (thousands). Not to mention the energy that people spend for it.
For the 2nd question – it’s a limited price deal, and it’ll go up in a few hours.
I use GPT-3 to generate synonyms for all of the emojis.
And yes, it’s simple to have a list of synonyms. But is it easy? Calculate the time anyone takes to write synonyms for a particular emoji, and multiply it by the number of emojis (thousands). Not to mention the energy that people spend for it.
For the 2nd question – it’s a limited price deal, and it’ll go up in a few hours.
> Calculate the time anyone takes to write synonyms
Did a quick Google search - there are ~3k emojis and ~100 were added in 2020.
Sure... it would've been a time consuming process to write synonyms for all emojis, but after the initial pass it shouldn't be a huge effort to update the list for new emojis. This would have the added benefit of having a human validated dataset, compared to what GPT3 might produce.
Also, there are resources such as Emojipedia which have at-least 2 or 3 synonyms per emoji. With that in perspective, the figure of 5k / 10k synonyms mentioned on your website doesn't seem that 'groundbreaking'.
I don't mean to undermine your work... the emoji search on macOS IS terrible. Just wanted to understand what value GPT3 brought to your product.
Did a quick Google search - there are ~3k emojis and ~100 were added in 2020.
Sure... it would've been a time consuming process to write synonyms for all emojis, but after the initial pass it shouldn't be a huge effort to update the list for new emojis. This would have the added benefit of having a human validated dataset, compared to what GPT3 might produce.
Also, there are resources such as Emojipedia which have at-least 2 or 3 synonyms per emoji. With that in perspective, the figure of 5k / 10k synonyms mentioned on your website doesn't seem that 'groundbreaking'.
I don't mean to undermine your work... the emoji search on macOS IS terrible. Just wanted to understand what value GPT3 brought to your product.
> With that in perspective, the figure of 5k / 10k synonyms mentioned on your website doesn't seem that 'groundbreaking'.
I only generated it once in a go. The number will grow as the time goes by. Imagine the multiplier then.
> Just wanted to understand what value GPT3 brought to your product.
Thanks for your intent, appreciate it :) So I used GPT-3 to train all emojis, not only the new ones. That means a particular emoji will get more and more synonyms as I hire GPT-3 over and over again.
I only generated it once in a go. The number will grow as the time goes by. Imagine the multiplier then.
> Just wanted to understand what value GPT3 brought to your product.
Thanks for your intent, appreciate it :) So I used GPT-3 to train all emojis, not only the new ones. That means a particular emoji will get more and more synonyms as I hire GPT-3 over and over again.
What are the system requirements? (For example, I’m on 10.9, I assume that won’t work.)
Hey! The support starts from Catalina (macOS 10.15) to the latest Big Sur. And works for Intel or M1 processor.
Thanks! You may want to add that to your website somewhere, especially since the requirements are a tad stringent (lots of stuff still supports Mojave, for instance).
Second question -- why does the pricing page say "182 spots left"? Is this a limited app release? I've never heard of anybody trying that before!