Mo Samsami (@m_r_samsami) 's Twitter Profile
Mo Samsami

@m_r_samsami

AI Researcher @Google @Mila_Quebec. Building agents capable of planning in complex situations.
Past: @ServiceNowRSRCH & @EPFL_en.

ID: 1419003842079952905

linkhttps://mrsamsami.github.io/ calendar_today24-07-2021 18:37:51

158 Tweet

443 Takipçi

511 Takip Edilen

Mo Samsami (@m_r_samsami) 's Twitter Profile Photo
Mo Samsami
@m_r_samsami

9 months ago

Larger models are more resilient to misinformation, thanks to their world model! 🌍 Introducing "Too Big to Fool: Resisting Deception in Language Models". Our paper shows that larger LLMs are less affected by deception and hypothesizes about the nature of this capability. 1/🧵

Larger models are more resilient to misinformation, thanks to their world model! 🌍 Introducing "Too Big to Fool: Resisting Deception in Language Models". Our paper shows that larger LLMs are less affected by deception and hypothesizes about the nature of this capability. 1/🧵