Study Guide > Changes.md

Changes.md University of Piraeus CIS NETWORKS This file includes all the changes that are required to run each of the models  ## Question 1 ### 1. 1 Layer RNN class model(nn.Module): def __init__(self,input_dim, embedding_dim, hidden_dim, output_dim): super(model, self).__init__() self.embedding_layer = nn.Embedding(num_embeddings=input_dim ...[Show More]

Changes.md

University of Piraeus
CIS NETWORKS
This file includes all the changes that are required to run each of the models

## Question 1
### 1. 1 Layer RNN
class model(nn.Module):
def __init__(self,input_dim, embedding_dim, hidden_dim, output_dim):
super(model, self).__init__()
self.embedding_layer = nn.Embedding(num_embeddings=input_dim,
embedding_dim=embedding_dim)
self.rnn = nn.RNN(input_size=embedding_dim, hidden_size=hidden_dim,
batch_first=True)
self.linear = nn.Linear(hidden_dim, output_dim)
def forward(self, X_batch):
embeddings = self.embedding_layer(X_batch)
output, hidden = self.rnn(embeddings)
logits = self.linear(output[:,-1])
probs = F.softmax(logits, dim=1)
return probs
### 2. 1 Layer Bidirectional RNN
class model(nn.Module):
def __init__(self,input_dim, embedding_dim, hidden_dim, output_dim):
super(model, self).__init__()
self.embedding_layer = nn.Embedding(num_embeddings=input_dim,
embedding_dim=embedding_dim)
self.rnn = nn.RNN(input_size=embedding_dim, hidden_size=hidden_dim,
batch_first=True, bidirectional=True)
self.linear = nn.Linear(hidden_dim*2, output_dim)
def forward(self, X_batch):
embeddings = self.embedding_layer(X_batch)
output, hidden = self.rnn(embeddings)
hidden_cat = torch.cat((hidden[-2, :, :], hidden[-1, :, :]), dim=1) #
concatenate final hidden states of forward and backward RNNs
logits = self.linear(hidden_cat)
probs = F.softmax(logits, dim=1)
return probs
### 3. 2 Layer Bidirectional RNN
class model(nn.Module):
def __init__(self,input_dim, embedding_dim, hidden_dim, output_dim):
super(model, self).__init__()
self.embedding_layer = nn.Embedding(num_embeddings=input_dim,
embedding_dim=embedding_dim)
self.rnn = nn.RNN(input_size=embedding_dim, hidden_size=hidden_dim,
batch_first=True, num_layers=2, bidirectional=True)
self.linear = nn.Linear(hidden_dim*2, output_dim)
def forward(self, X_batch):
embeddings = self.embedding_layer(X_batch)
output, hidden = self.rnn(embeddings)
hidden_cat = torch.cat((hidden[-2, :, :], hidden[-1, :, :]), dim=1)
logits = self.linear(hidden_cat)
probs = F.softmax(logits, dim=1)
return probs