We find that LLMs prioritize their internal expectations over conflicting visual evidence.

On AI agents, grunt work, and the part of science that isn't replaceable.

This PR aims to fix #4066. Briefly, it sets names for the layers part of a plot that should be able to uniquely identify a layer. They're specified in the params argument to layer() and default...

This PR aims to fix #5178. Briefly, it adds labs(dict) that allows one to use a data-dictionary to label a plot based on variable names (rather than aesthetics). Let's just jump into examples. ...

This PR aims to (eventually) fix #3062. Briefly, it adds a layer(layout) argument to control how data is assigned to panels. Currently only layout = "fixed" is implemented, which repeats ...

This PR aims to fix #3501 and replaces #6066. It improves upon #6066 by simplifying the stat to stat_manual(fun = identity). It really is a thin wrapper to essentially offer a way for a user to pro...

This PR aims to fix #6020. Briefly, it allows expand to take a length-4 logical vector for each of the positions (top, right, bottom, left). Alternatively, one could use a named vector too. devtool...