Nicolas Gambardella<p>Interesting: one can train deep nets without <a href="https://genomic.social/tags/normalization" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>normalization</span></a> layers by replacing them with a parameterized tanh()<br><a href="https://arxiv.org/abs/2503.10622" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="">arxiv.org/abs/2503.10622</span><span class="invisible"></span></a><br>tanh() are my favourite <a href="https://genomic.social/tags/activation" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>activation</span></a> function<br><a href="https://genomic.social/tags/deeplearning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>deeplearning</span></a> <a href="https://genomic.social/tags/artificialintelligence" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>artificialintelligence</span></a></p>