Language Models Are Capable of Metacognitive Monitoring and Control of Their Internal Activations

Devs

Language Models Are Capable of Metacognitive Monitoring and Control of Their Internal Activations | Read Paper on Bytez