Update main.py

main.py

@@ -10,6 +10,210 @@ from collections import OrderedDict
 
 app = Flask(__name__)
 
+
+device = torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')
+if device == 'cuda:0':
+  torch.cuda.set_device(device)
+print(device)
+
+def extract_text_from_link(url):
+  response = requests.get(url)
+  soup = BeautifulSoup(response.content, 'html.parser')
+  text = soup.get_text()
+  return text
+
+
+
+doc = """The word "deep" in "deep learning" refers to the number of layers through which the data is transformed. More precisely,
+deep learning systems have a substantial credit assignment path (CAP) depth. The CAP is the chain of transformations from input to
+output. CAPs describe potentially causal connections between input and output. For a feedforward neural network, the depth of the
+CAPs is that of the network and is the number of hidden layers plus one (as the output layer is also parameterized). For recurrent
+neural networks, in which a signal may propagate through a layer more than once, the CAP depth is potentially unlimited.[13] No
+universally agreed-upon threshold of depth divides shallow learning from deep learning, but most researchers agree that deep
+learning involves CAP depth higher than 2. CAP of depth 2 has been shown to be a universal approximator in the sense that it
+can emulate any function.[14] Beyond that, more layers do not add to the function approximator ability of the network. Deep
+models (CAP > 2) are able to extract better features than shallow models and hence, extra layers help in learning the features
+effectively."""
+
+
+class Text2Words:
+  def __init__(self, document):
+    self.text_all = re.findall(r'\b[A-Za-z]+\b', document)
+    self.text = list(set(self.text_all))
+    self.chars_all = ''.join(self.text)
+    self.chars = self.unique_chars(self.chars_all)
+    self.int2char = dict(enumerate(self.chars))
+    self.char2int = {char: ind for ind, char in self.int2char.items()}
+    self.maxlen = len(max(self.text, key=len))
+    self.update_text()
+    self.input_seq_char, self.target_seq_char = self.get_seq_char(self.text)
+    self.input_seq_index, self.target_seq_index = self.get_seq(self.char2int, self.input_seq_char, self.target_seq_char, len(self.text))
+    self.dict_size = len(self.char2int)
+    self.seq_len = self.maxlen - 1
+    self.batch_size = len(self.text)
+    self.input_seq = self.one_hot_encode(self.input_seq_index, self.dict_size, self.seq_len, self.batch_size)
+
+  def one_hot_encode(self, sequence, dict_size, seq_len, batch_size):
+    # Creating a multi-dimensional array of zeros with the desired output shape
+    features = np.zeros((batch_size, seq_len, dict_size), dtype=np.float32)
+
+    # Replacing the 0 at the relevant character index with a 1 to represent that character
+    for i in range(batch_size):
+        for u in range(seq_len):
+            features[i, u, sequence[i][u]] = 1
+    return features
+
+  def get_seq(self, char2int, input_seq_char, target_seq_char,n):
+    x=[]
+    y=[]
+    for i in range(n):
+      x.append([char2int[character] for character in input_seq_char[i]])
+      y.append([char2int[character] for character in target_seq_char[i]])
+    return x,y
+
+  def get_seq_char(self, text):
+    input_seq = []
+    target_seq = []
+
+    for i in range(len(text)):
+        # Remove last character for input sequence
+      input_seq.append(text[i][:-1])
+        # Remove first character for target sequence
+      target_seq.append(text[i][1:])
+    return input_seq, target_seq
+
+  def unique_chars(self, chars_all):
+    chars = []
+    for letter in chars_all:
+      if letter not in chars:
+        chars.append(letter)
+    # chars = sorted(chars)
+    if ' ' not in chars:
+      chars.append(' ')
+    return sorted(chars)
+
+  def update_text(self):
+    for i in range(len(self.text)):
+      while len(self.text[i])<self.maxlen:
+          self.text[i] += ' '
+
+  def description(self):
+    text = {}
+    for word in self.text:
+      char = word[0]
+      if char not in text:
+        text[char] = []
+      text[char].append(word.strip())
+    for k,v in (sorted(text.items())):
+      print(f'{k} : {sorted(v)}')
+
+  def lengt_analysis(self):
+    text = {}
+    words = set(self.text_all)
+    for word in words:
+      n = len(word)
+      if n not in text:
+        text[n] = []
+      text[n].append(word.strip())
+    for k,v in (sorted(text.items())):
+      print(f'{k} : count = {len(v)} list = {sorted(v)}')
+    return None # text
+
+
+def create_object(doc):
+  return Text2Words(doc)
+
+
+def get_inputs(obj):
+  input_seq = torch.tensor(obj.input_seq, device=device)
+  target_seq_index = torch.tensor(obj.target_seq_index, device=device)
+  return input_seq, target_seq_index
+
+class Model(nn.Module):
+    def __init__(self, input_size, output_size, hidden_dim, n_layers):
+        super(Model, self).__init__()
+
+        # Defining some parameters
+        self.hidden_dim = hidden_dim
+        self.n_layers = n_layers
+
+        #Defining the layers
+        # RNN Layer
+        self.rnn = nn.RNN(input_size, hidden_dim, n_layers, batch_first=True)
+        # Fully connected layer
+        self.fc = nn.Linear(hidden_dim, output_size)
+
+    def forward(self, x):
+        batch_size = x.size(0)
+        hidden = self.init_hidden(batch_size)
+        out, hidden = self.rnn(x, hidden)
+        out = out.contiguous().view(-1, self.hidden_dim)
+        out = self.fc(out)
+        return out, hidden
+
+    def init_hidden(self, batch_size):
+        # This method generates the first hidden state of zeros
+        torch.manual_seed(42)
+        hidden = torch.zeros((self.n_layers, batch_size, self.hidden_dim), device=device)
+        return hidden
+
+def create_model(obj):
+  model = Model(input_size=obj.dict_size, output_size=obj.dict_size, hidden_dim=2*obj.dict_size, n_layers=1)
+  model.to(device)
+  lr=0.01
+  criterion = nn.CrossEntropyLoss()
+  optimizer = torch.optim.Adam(model.parameters(), lr=lr)
+  return model, criterion, optimizer
+
+# This function takes in the model and character as arguments and returns the next character prediction and hidden state
+def predict(model, character):
+    # One-hot encoding our input to fit into the model
+    # print(character)
+    character = np.array([[obj.char2int[c] for c in character]])
+    # print(character)
+    character = obj.one_hot_encode(character, obj.dict_size, character.shape[1], 1)
+    # print(character,character.shape)
+    character = torch.tensor(character, device=device)
+    character.to(device)
+    out, hidden = model(character)
+    # print(out, hidden)
+    prob = nn.functional.softmax(out[-1], dim=0).data
+    # print(prob)
+    char_ind = torch.max(prob, dim=0)[1].item()
+    # print(sorted(prob, reverse=True))
+    return obj.int2char[char_ind], hidden
+
+# This function takes the desired output length and input characters as arguments, returning the produced sentence
+def sample(model, out_len, start='h'):
+    model.eval() # eval mode
+    chars = [ch for ch in start]
+    char = chars[-1]
+    chars = chars[:-1]
+    # Now pass in the previous characters and get a new one
+    while char != ' ':
+        chars.append(char)
+        char, h = predict(model, chars)
+    return ''.join(chars)
+
+
+def load_checkpoint(filepath):
+    checkpoint = torch.load(filepath)
+    # print(checkpoint['state_dict'])
+    model = checkpoint['model']
+    # print(model)
+    model.load_state_dict(checkpoint['state_dict'])
+    # print(model.parameters())
+    # for parameter in model.parameters():
+    #     parameter.requires_grad = False
+        # print(parameter)
+
+
+    model.eval()
+    return model
+
+
 @app.route('/')
 def home():
-	return {'key':'Hello HuggingFace!'}
+    model = load_checkpoint('checkpoint.pth')
+    res = sample(model, obj.maxlen, 'ap')
+	return {'key':res}