PinnedSushil KhadkainTowards AICreativity in Language ModelsThe Science of Creativity in LLMs: Softmax and Temperature ExplainedJun 181Jun 181
PinnedSushil KhadkainTowards AIInformation & EntropyWhat, Why, and How explainedAug 15, 2023Aug 15, 2023
PinnedSushil KhadkainTowards AISelf-Attention in TransformersA Beginner-Friendly Guide to Self-Attention MechanismMay 15, 20235May 15, 20235
PinnedSushil KhadkaMixture of Gaussians (MoG) for Background/Foreground Segmentation — Part 1Have you ever wondered how people used to detect moving objects before the deep learning era?Feb 15, 20231Feb 15, 20231
Sushil KhadkaThanks for the post, however, I can't entirely agree here.Each pixel's distribution is not modeled, if this was the case then we would have a one-dimensional probability distribution for each…Feb 7Feb 7
Sushil KhadkaIn PatchEmbedding class, you randomly initialized cls_token with shape (1, 1, emb_size), and in the…Is it because the batch_size information is not available inside the init function or is it because you want the same values during…Jul 9, 20231Jul 9, 20231