this is an artistic research project looking into how large datasets for pretraining LLMs "clean" their crawled data, and how we can use these heuristics get our data too dirty to be trained...
@#$%^&*()!$%^&*)(*&^!@#$%(*$%^*@#$%)(*$%^@#$(*&#$%!@#$)(*&^%$##$%%^&**&^#$^%@#$%*&#$%^&**$%#$%^^%$%^%*&**&^%@@#$%$#@#$%^%$ER^*(&^*(*&^^&(*&*()(*&^%^&*&^%$#^%$#$(*&^%
Updated 2026年01月09日 21:33:07 +01:00
a lil old web project, from 2021, that use the web's structure as the structure for showing contents
Updated 2026年01月07日 20:21:13 +01:00