"LazImpa": Lazy and Impatient neural agents learn to communicate efficiently

Crucially, ZLA is considered to be an efficient property Previous work has shown that artificial of our language (Gibson et al., 2019). Besides the neural agents naturally develop surprisingly obvious fact that an efficient code would be easier non-efficient codes. This is illustrated to process for us, it is also argued to be a core property by the fact that in a referential of natural language, likely to be correlated game involving a speaker and a listener with other fundamental aspects of human communication, neural networks optimizing accurate transmission such as regularity and compositionality over a discrete channel, the emergent (Kirby, 2001). Encouraging it might hence lead messages fail to achieve an optimal to emergent languages that are also more likely to length. Furthermore, frequent messages develop these other desirable properties.