1/4: The lottery ticket hypothesis suggests that by training DNNs from “lucky” initializations, we can train networks which are 10-100x smaller with minimal performance losses. In new work, we extend our understanding of this phenomenon in several ways... https://ai.facebook.com/blog/understanding-the-generalization-of-lottery-tickets-in-neural-networks …pic.twitter.com/ZXGWjBWjcz
-
-
3/4: Is the lottery ticket phenomenon a general property of DNNs or merely an artifact of supervised image classification? We show that the lottery ticket phenomenon is a general property which is present in both
#reinforcementlearning and#NLP: https://arxiv.org/abs/1906.02768?fbclid=IwAR2prqiChhwXj8_CqxyMcV1ePqVzFvojYgsuieRmP94xE8Ewa5kAkI_T1_U …Prikaži ovu nit -
4/4: Can we begin to explain lottery tickets theoretically? We introduce a new theoretical framework on the formation of lottery tickets to help researchers advance toward a better understanding of lucky initializations: https://arxiv.org/pdf/1905.13405.pdf?fbclid=IwAR0MKAue_0k1d-DugT8UBAvRM_fjKBodtQsSqNel1gGF8EVmV14aChCCBMA …
Prikaži ovu nit
Kraj razgovora
Novi razgovor -
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.