The Rise of the Transformer Architecture

 Todays  presentation is the transformer lecture from the Berkeley full stack deep learning course (which is quite excellent).  They title it 'Transfer Learning and Transformers', but the transfer learning part is really just a few minutes at the very beginning that is really just a lead in to their view of the events leading up to transformer architectures.

So we quickly mention transfer learning, then move on to vector embedding models of language, then on to the real meat of the lecture, the transformer architecture.  We recently covered the OpenAI Codex automated coding system, and they talk about how that fell out of this transformer research on natural language modeling.  They also briefly talk about ethic and ai systems like GPT-3 or Codex.

This lecture is from the spring 2021 Full Stack Deep Learning online course.


HTC has a number of blog posts on transformers you can check out here, including a full tutorial on how to code one in PyTorch.

We also have covered transfer learning before if you want more info on that in posts as well as in the HTC deep learning courses which dive into how to code it in PyTorch.


Observations

1:  The lecture mentions at one point how architectures prior to transformers didn't really consider the original word relationships except in the first layer.  Which got me thinking about U-Nets, which is the magic architecture for image processing, and what the NLP equivalent would be for that architecture.  We;ll leave that thought for the reader to puzzle out.

Comments

Post a Comment

Popular posts from this blog

CycleGAN: a GAN architecture for learning unpaired image to image transformations

Image
 Like the Doublemint Twins touting the joys of Doublemint gum, 2 GANs are surely better than 1 GAN.  Especially if we package them together inside of one meta GAN module.  And this is exactly what the CycleGAN architecture does. Have you ever harbored dark secrets of turning a horse into a zebra?  The CycleGAN was developed to do just that. Learn how to turn a horse into a zebra.  And more. Now right away you can notice a difference between the image to image transformation GAN architectures we've been discussing over the last few posts.  Those last few posts described systems that learn from a database of matched input-output image pairs. And if your goal is to turn an edge representation into a nicely filled in continuous tone image, it's easy to build your database of matched input-output image pairs that your GAN system can then learn off of.  Take a continuous tone image (which will be the output of the database pair entry), then run it through an edge detector algorithm. 
Read more

Pix2Pix: a GAN architecture for image to image transformation

Image
 I thought following up yesterday's TraVelGAN post with a Pix2Pix GAN post would be useful to compare what is going on in the 2 architectures.  Two different approaches to the same problem. I stole this Pix2Pix Overview slide below from an excellent deeplearning.ai GAN course (note that they borrowed it from the original paper) because it gives you a good feel for what is going on inside of the Pix2Pix architecture.    Note how the Generator part is very much like an auto-encoder architecture, but rebuilt using the U-Net architecture features (based on skip-connections) that fastai has been discussing in their courses for several years before it became more widely known to the deep learning community at large  (and which originally came from an obscure medical image segmentation paper) . So the Generator in this Pix2Pix GAN is really pretty sophisticated, consisting of a whole image to image auto-encoder network with U-Net skip connections to generate better image quality at highe
Read more

Smart Fabrics

Image
"What makes smart fabrics revolutionary is that they have the ability to do many things that traditional fabrics cannot, including communicate, transform, conduct energy and even grow."  This is a quote from Rebeccah Pailes-Friedman, who is a professor at Pratt with a research focus on developing wearable technology and smart textiles. Enhancing textiles can be viewed through different lens.  Enhancements can be aesthetic, or performance boosting, or durability boosting. 
Read more