Hi all, I published a summary of a recent paper published by researchers from MIT, University of Toronto and Cambridge. It takes a deeper look into the Lottery Ticket Hypothesis, an idea that neural networks have a pruned version which can train just as well as the original version with significantly less size, when applying the right weights (this is a “winning ticket”). The result could help better understand “winning tickets” and is generally interesting. I hope you’ll like it and happy to get feedback. Full summary here: https://www.lyrn.ai/2019/07/02/exploring-the-lottery-ticket-hypothesis/