¡Hola, Mundo!

RNN model Used for sequential data modeling. (A-B-C-D-?, ? is probably E considered of ahead) Special types of fully connected neural network. Connection weights are shared by all layers. Encoder It recieves all the words sequentially, and at the end, it compresses all of these word into A Vector which is called a CONTEXT VECTOR. context vector It recieves all the words and then passed the hidde..

Word2Vec Steps 1. Convert words to one-hot-vector 2. Get W(lookup table) : V * M (V : len_of_words * M : Hyperparameter, usually used square of 2) Neural networks Emedding Steps 1. Conver words to integer 2. Get embedding table(lookup table) : num_embeddings * embedding_dim (num_embeddings : len_of_words * embedding_dim : Hyperparameter) Reference : https://wikidocs.net/60854 https://wikidocs.ne..

randint randint(start, end, (tensor size)) torch.randint(3, 5, (3,)) >>> tensor([4, 3, 4]) torch.randint(3, 10, (2, 2)) >>> tensor([[4, 5], [6, 7]]) scatter_ scatter_(dim, index, value) y.unsqueeze(1) >>> tensor([[3], [1], [2]]) y_one_hot.scatter_(1, y.unsqueeze(1),1) >>> tensor([[0., 0., 0., 1., 0.], # tensor(1.7024, grad_fn=) - cost function 2 : log_softmax (y_one_hot * - F.log_softmax(model(x..

Step 1. Initialize weight and bias : requires_grad=True (if False, This will prevent updating of model weights during fine-tuning.) 2. Set the optimizer and learning rate 3. Set hypothesis 4. Get cost 5. Initialize optimizer : zero_grad() 6. backward : differenciate cost function -> get gradient -> update w, b when backpropagation 7. step : apply learning rate to w, b -> update w, b optimizer(di..

In 3-dimension tensor, count how many items are there. 0th dimension [ [ [ , ] ] ] [] = 3 1st dimension [ [ [ , ] ] ] [] = 1 2nd dimension [ [ [ , ] ] ] [] = 2 tensor([[[0, 1]], [[1, 1]], [[2, 1]]]) torch.Size([3, 1, 2]) FloatTensor Create matrix. torch.FloatTensor([[1.,2.],[3.,4.],[5.,6.]]) >>> tensor([[1., 2.], [3., 4.], [5., 6.]]) view Same function as reshape in numpy. -1 : It means I am not..