@poleguy The paper I linked downthread tries several approaches! It does not look like (in the current models) a description of the language is good enough. An interactive session does help somewhat in their experiment.
I would like to see a non-Esolang example, although the effect seems very strong there.
While (b) is certainly a questionable assumption, I think it is currently borne out by this and similar results -- the models get good by having training sets that are relevant, rather than developing transferrable knowledge.
In reply to
Mark Gritter
@markgritter@mathstodon.xyz
Software Engineer at Thirdlaw. Previously co-founded Tintri, on Vault team at HashiCorp, founding engineer at Akita Software, Principal Engineer at Postman. Big nerd. he/him
mathstodon.xyz
Mark Gritter
@markgritter@mathstodon.xyz
Software Engineer at Thirdlaw. Previously co-founded Tintri, on Vault team at HashiCorp, founding engineer at Akita Software, Principal Engineer at Postman. Big nerd. he/him
mathstodon.xyz
@markgritter@mathstodon.xyz
·
Apr 14, 2026
1
0
0
Loading comments...