Training large language models on narrow tasks can lead to broad...

ha condiviso un link

2026-01-14 20:16:24 - Translate -

WWW.NATURE.COM

Training large language models on narrow tasks can lead to broad misalignment

Nature, Published online: 14 January 2026; doi:10.1038/s41586-025-09937-5Finetuning a large language model on a narrow task of writing insecure code causes a broad range of concerning behaviours unrelated to coding.

0 Commenti 0 condivisioni 31 Views 0 Anteprima

{if $system['cookie_consent_enabled']} {/if}