/explore
|
You are here 
|
evanfields.net | ||
| | | | |
ezyang.github.io
The Rule of Three in software says that you should be willing to duplicate a piece of code once, but on the third copy you should refactor. This is a refinement on DRY (Don't Repeat Yourself) accounting for the fact that it might not necessarily be obvious how to eliminate a duplication, and waiting until the third occurrence might clarify. (See also The Wrong Abstraction.) LLMs love to duplicate code. Think about how, absent any other prompting, if you ask an LLM to modify a program, it will spit out a new copy of the program with the changes requested. For an LLM to decide to refactor code to remove duplication requires a moment where the LLM proactively decides to go out of their way to do some cleanup. (Actually, Sonnet 3.7 also has some of this inclinat... |
|
| | | | | ||
| | | | |
simonwillison.net
Anthropic released a new model this morning, and I think it's likely now the single best available LLM. Claude 3 Opus was already mostly on-par with GPT-4o, and the new ... |
|
| | | | | ||
| | | | |
mattsayar.com
Not that long ago, I tried to use Claude to customize Publii, the app I use to make this website. It didn't do so well. Literally the next day, Anthropic announced the Claude 4 family, and now Claude Sonnet 4.5 is out. Anthropic's unbiased opinion... |
|
| | | | | ||
| | | | |
www.destroyallsoftware.com
|
|
| | | |||
