GustavoLucasCarvalho

Working on ai safety research with @johntzwei and @robinomial / @NorthwesternU & @CSatUSC alum

[Johnny Tian-Zheng Wei @johntzwei]

Hi all, surreal moment: the recommendations in our social media regulation paper have been adopted by EU law!

The Digital Services Act previously required companies to report the "accuracy" of content moderation systems, and we suggested precision and recall instead.🧵[1/4]

[GustavoLucasCarvalho @CtrlVGustavo]

Check out our work on the interesting connections between LoRA, conditional steering, and reward modeling! See yall in sf #COLM2026 🔥🔥🔥

[Zitong Huang @ZitongHuang12]

Can training a language model for binary classification improve generation?
Surprisingly, yes.

We introduce Token-Level Off-Policy Labeling (TOPL) for faithful generation under distribution shift. Instead of optimizing next-token prediction, we train an LLM to classify whether each generated token is faithful.

🔗 arXiv: https://t.co/LpADmQFvWl
🧵👇

[GustavoLucasCarvalho @CtrlVGustavo]

@DeqingFu That’s the paper that made me apply to usc haha. So cool to see it getting recognition :)

[GustavoLucasCarvalho @CtrlVGustavo]

@lexfridman In that analogy, who’s the one holding the carrot in front of us with a cartoonish fishing rod and smacking us with the stick from behind?

[GustavoLucasCarvalho @CtrlVGustavo]

@Jammalz1_ vc é significativo para mim seu boçal