Attention is Not All You Need - the transformer backlash begins

 With the end of my 45 day straight 16 hour a day campout in the debugger doing stack traces for a major software release, i finally have some time to get back to this blog.  And in the fast paced world of deep learning, there are a ton of new things to cover.

A paper was just release called 'Attention is not all you need: pure attention looses rank doubly with depth'.  It tries to dive into answering the question 'why do transformers work'.  And it turns out our old friend the skip connection plays the key role (surprise).


Here's the abstract.

Attention-based architectures have become ubiquitous in machine learning. Yet our understanding of the reasons for their effectiveness remains limited. This work proposes a new way to understand self-attention networks: we show that their output can be decomposed into a sum of smaller terms, each involving the operation of a sequence of attention heads across layers. Using this decomposition, we prove that self-attention possesses a strong inductive bias towards “token uniformity”. Specifically, without skip connections or multi-layer perceptrons (MLPs), the output converges doubly exponentially to a rank-1 matrix. On the other hand, skip connections and MLPs stop the output from degeneration. Our experiments verify the identified convergence phenomena on different variants of standard transformer architectures.


Another fascinating quote.

Skip connections were first introduced in ResNets, ever since, it has been used to facilitate optimization in deep networks . In particular, skip connections tackle the vanishing gradient problem, by allowing the gradient to flow bypass the skipped layers during backpropagation. The original motivation of using skip connections in transformers follow the same reasoning on facilitating optimization . With the paths decomposition for transformers, we discover an additional surprising importance of skip connections: they prevent the transformer output from degenerating to rank one exponentially quickly with respect to network depth.

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