Simplified interface for doing Deep Learning in TensorFlow (mimicking Scikit Learn)
I was working with bert to encode strings, used uncased 768 A-12 model. I guess its not performing well, I mean, look at what I found...
x = "The concept of space is considered to be of fundamental importance to an understanding of the physical universe"
y = "Agriculture is the science and art of cultivating plants and livestock"
These two sentence has cosine score of [[0.7106109]]
I mean, is is something wrong with the model, or something I am doing wrong?
good first issue
and start contributing to them. It is the best and fastest way to familiarize yourself with the codebase! Goodluck :)
Hey @akashnd look into these links, if you could get started with working on any of these issues (whichever you feel would be a good fit for you)
@vishwanath63 @rishishounak @Sheikh-Inzamam @kritikaparmar-programmer @Sudheeradh
Hey @akashnd look into these links, if you could get started with working on any of these issues (whichever you feel would be a good fit for you)
@vishwanath63 @rishishounak @Sheikh-Inzamam @kritikaparmar-programmer @Sudheeradh
hey @Aryaman2912, the first step would be brush up your basic deeplearning concepts ( have knowledge of DNN, CNN, RNN, LSTM ).
after that, I would suggest you to start the tensorflow-in-practice course available on deeplearning.ai (COURSERA).
I've done that, made few projects, and worked on modeling a few anime GANs.