speaking of transformers—when my professor for Deep Learning was introducing them to us (working up from the intuition of what could arguably be improved from RNNs/GRUs/LSTMs/Seq2Seq), he paused for a moment to say:
….and what this resulted in is a paper in 2017 called “Attention Is All You Need”, that introduced an architecture for machine translation called the Transformer model.
As a side note, I think it's a terrible name for a neural network architecture, because basically every neural net is a transformer... it transforms an input to an output. I don't know why this one gets to claim that name in particular. Anyway. That's what it's called :/
(the delivery is missing, but you could hear the mix of…. professional yet very clear disappointment / frustration / resentment at the name. especially the “I don't know why this one gets to claim that name in particular” bit)
Generative AI can change the game for the retail industry It is a branch of artificial intelligence that can create new content, such as images, text, music, and more. It has many applications in retail, such as personalizing product recommendations, generating realistic product images, creating engaging marketing content, and optimizing inventory management. Generative AI can transform customer experiences and business efficiency in various ways. Moreover, there are several advantages and challenges of generative AI for retail to consider and it is essential to understand how you can use it for your own business.
To read the blog and learn more about generative AI in retail https://www.webcluesinfotech.com/generative-ai-in-retail-transforming-customer-experiences-business-efficiency
I present to your attention Soundwave from "Cyberverse". The model is movable, as a demonstration, I decided to recreate several of his poses from the series.
Abstract
Deep neural networks (DNNs) are often used for text classification due to their high accuracy. However, DNNs can be computationally intensive, requiring millions of parameters and large amounts of labeled data, which can make them expensive to use, to optimize, and to transfer to out-of-distribution (OOD) cases in practice. In this paper, we propose a non-parametric alternative to DNNs that’s easy, lightweight, and universal in text classification: a combination of a simple compressor like gzip with a k-nearest-neighbor classifier. Without any training parameters, our method achieves results that are competitive with non-pretrained deep learning methods on six in-distribution datasets. It even outperforms BERT on all five OOD datasets, including four low-resource languages. Our method also excels in the few-shot setting, where labeled data are too scarce to train DNNs effectively. Code is available at https://github.com/bazingagin/npc_gzip.
(July 2023 – pdf)
“this paper's nuts. for sentence classification on out-of-domain datasets, all neural (Transformer or not) approaches lose to good old kNN on representations generated by.... gzip” [x]