Why Hacker News recommends it
5 comments from 3 distinct recommending readers, between 2023 and 2024.
- Recommends
This post from @gfodor is also a really interesting read on the topic of compressing prompts, not sure what the main use case would be though! twitter.com/gfodor/status/1643444605332…
Increase GPT4's context size by asking it to compress your promptsRead the thread - Recommends
Another interesting tweet[1] I saw today shows how you can ask ChatGPT to compress text and it invents it's own (effective!) shorthand. I bet it's related somehow to glitch tokens and the way GPT is grouping tokens internally. [1] mobile.twitter.com/VictorTaelin/status…...
OpenAI TokenizerRead the thread - Recommends
I spent some time on this, here is a better version: mobile.twitter.com/gfodor/status/164341…
Increase GPT4's context size by asking it to compress your promptsRead the thread - Cites it
There was some chatter about 9 months ago I think when folks were using compression, unicode and all sorts of weird hacks to increase the context window. This is rather recent, but in the same vein. microsoft.com/en-us/research/blog/llmli…... twitter.com/VictorTaelin/status/1642664…
A woman bought a vintage dress. It had a secret pocket with a mysterious noteRead the thread - Disagrees
One reply points out that even though the text length is 1/4 of the original, the “compressed” output is actually LONGER in token count! twitter.com/dogeofcoin/status/164291889…
Increase GPT4's context size by asking it to compress your promptsRead the thread