Elektrine lite

← Feed

Bruno Gavranović

bgavran@mathstodon.xyz

<p>I&#39;m building neural networks that generate provably correct code, and the software infrastructure for training them.</p><p>Recently experimenting with TensorType: <a href="https://github.com/bgavran/TensorType" target="_blank" rel="nofollow noopener" translate="no"><span class="invisible">https://</span><span class="">github.com/bgavran/TensorType</span><span class="invisible"></span></a></p>

Posts

  • Post #1506834

    I never wrote about it here, but as of some time ago I figured out a basic implementation for named axes in TensorType: https://github.com/bgavran/TensorType This means that now you&amp;#39;re: a) forced to assign some meaning to all your axes b) cannot by accident sum over sequence length, for instance, instead of &amp;quot;batch&amp;quot; There&amp;#39;s still a long way to go to get this fully integrated, but I&amp;#39;m quite excited

  • Post #1506825

    I just published a new blog post! Types and Neural Networks (https://www.brunogavranovic.com/posts/2026-04-20-types-and-neural-networks.html) It&amp;#39;s about what comes into view once dependent type systems become a part of neural network training.