Do Large Language Models Perform Latent Reasoning?

Explore the intriguing question of whether large language models engage in latent reasoning in this remote talk by Mor Geva from Tel Aviv University. Delve into the computational model of transformers and their potential for complex cognitive processes. Gain insights into cutting-edge research on artificial intelligence and natural language processing as part of the "Transformers as a Computational Model" series presented by the Simons Institute. In just 32 minutes, discover the latest findings and theories surrounding the cognitive capabilities of advanced language models.

Do Large Language Models Perform Latent Reasoning? (Remote Talk)