inputs = [] outputs = [] ''' create a one hot encoded tensor for an index. that is, the number i is mapped to a tensor of length n, where n is the number of tokens, filled with zeros and with a 1 in the ith index ''' def create_tensor(index): global embeddings n = len(embeddings) tensor = torch.zeros(n) tensor[index] = 1.0 return tensor for index in range(len(tokens)-1): token = tokens[index] embedding = map[token] next_token = tokens[index+1] next_embedding = map[next_token] inputs.append(embedding['embedding']) outputs.append(create_tensor(next_embedding['index'])) pass

''' construct the model ''' network = nn.Sequential( nn.Linear(20, 20), nn.Tanh(), nn.Linear(20,len(embeddings)), nn.Sigmoid() ) opt = op.SGD(network.parameters(), lr=0.01) err = nn.CrossEntropyLoss() index = 1 def callback(current_loss): global index print('run = '+str(index)+' and current error is '+str(current_loss.item()) ) index += 1 pass #torch.stack will convert the list of tensors to a tensor tc.train(network, opt, err, torch.tensor(inputs), torch.stack(outputs, dim=0), 3, callback=callback) torch.save(network.state_dict(), './text/model.txt')