update roberta script

.ipynb_checkpoints/Untitled-checkpoint.ipynb

@@ -0,0 +1,6 @@
+{
+ "cells": [],
+ "metadata": {},
+ "nbformat": 4,
+ "nbformat_minor": 4
+}

.ipynb_checkpoints/Untitled1-checkpoint.ipynb

@@ -0,0 +1,6 @@
+{
+ "cells": [],
+ "metadata": {},
+ "nbformat": 4,
+ "nbformat_minor": 4
+}

.ipynb_checkpoints/logistic-regression-checkpoint.ipynb

@@ -0,0 +1,6 @@
+{
+ "cells": [],
+ "metadata": {},
+ "nbformat": 4,
+ "nbformat_minor": 4
+}

logistic-regression.ipynb

@@ -0,0 +1,103 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 34,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pandas as pd\n",
+    "import numpy as np\n",
+    "import torch\n",
+    "import gensim.downloader as gn\n",
+    "import csv\n",
+    "from nltk.tokenize import word_tokenize"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 36,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "STEP 3 - PREPROCESSING\n"
+     ]
+    }
+   ],
+   "source": [
+    "names = ['content', 'id', 'label']\n",
+    "train_data_content = pd.read_table('train/in.tsv', error_bad_lines = False, header = None, quoting = csv.QUOTE_NONE, names = ['content', 'id'])\n",
+    "train_data_labels = pd.read_table('train/expected.tsv', error_bad_lines = False, header = None, quoting=csv.QUOTE_NONE, names = ['label'])\n",
+    "dev_data = pd.read_table('dev-0/in.tsv', error_bad_lines = False, header = None, quoting = csv.QUOTE_NONE, names = ['content', 'id'])\n",
+    "test_data = pd.read_table('test-A/in.tsv', error_bad_lines = False, header = None, quoting = csv.QUOTE_NONE, names = ['content', 'id'])\n",
+    "\n",
+    "print('STEP 3 - PREPROCESSING')\n",
+    "# lowercase all content\n",
+    "X_train = train_data_content['content'].str.lower()\n",
+    "y_train = train_data_labels['label']\n",
+    "X_dev = dev_data['content'].str.lower()\n",
+    "X_test = test_data['content'].str.lower()\n",
+    "\n",
+    "# tokenize datasets\n",
+    "X_train = [word_tokenize(content) for content in X_train]\n",
+    "X_dev = [word_tokenize(content) for content in X_dev]\n",
+    "X_test = [word_tokenize(content) for content in X_test]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 37,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "[==================================================] 100.0% 1662.8/1662.8MB downloaded\n"
+     ]
+    }
+   ],
+   "source": [
+    "w2v = gn.load('word2vec-google-news-300')\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.8.5"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 4
+}

logistic-regression.py

@@ -0,0 +1,123 @@
+import pandas as pd
+import numpy as np
+import csv
+import torch
+from nltk.tokenize import word_tokenize
+from gensim import downloader
+
+FEATURES = ['content', 'id', 'label']
+PATHS = ['train/in.tsv', 'train/expected.tsv', 'dev-0/in.tsv', 'test-A/in.tsv', './dev-0/out.tsv', './test-A/out.tsv']
+PRE_TRAINED = 'word2vec-google-news-300'
+
+class NeuralNetwork(torch.nn.Module):
+    def __init__(self, INPUT_DIM):
+        super(NeuralNetwork, self).__init__()
+        self.l1 = torch.nn.Linear(INPUT_DIM, 500)
+        self.l2 = torch.nn.Linear(500, 1)
+
+    def forward(self, x):
+        x = self.l1(x)
+        x = torch.relu(x)
+        x = self.l2(x)
+        x = torch.sigmoid(x)
+        return x
+
+def get_data(FEATURES, PATHS):
+  x_train = pd.read_table(PATHS[0], error_bad_lines = False, header = None, quoting = csv.QUOTE_NONE, names = FEATURES[:2])
+  y_train = pd.read_table(PATHS[1], error_bad_lines = False, header = None, quoting = csv.QUOTE_NONE, names = FEATURES[2:])
+  x_dev = pd.read_table(PATHS[2], error_bad_lines = False, header = None, quoting = csv.QUOTE_NONE, names = FEATURES[:2])
+  x_test = pd.read_table(PATHS[3], error_bad_lines = False, header = None, quoting = csv.QUOTE_NONE, names = FEATURES[:2])
+
+  return x_train, y_train, x_dev, x_test
+
+def preprocess(x_train, y_train, x_dev, x_test):
+  x_train = x_train[FEATURES[0]].str.lower()
+  x_dev = x_dev[FEATURES[0]].str.lower()
+  x_test = x_test[FEATURES[0]].str.lower()
+  y_train = y_train[FEATURES[2]]
+
+  return x_train, y_train, x_dev, x_test
+
+def tokenize(x_train, x_dev, x_test):
+  x_train = [word_tokenize(i) for i in x_train]
+  x_dev = [word_tokenize(i) for i in x_dev]
+  x_test = [word_tokenize(i) for i in x_test]
+
+  return x_train, x_dev, x_test
+
+def use_word2vec():
+  w2v = downloader.load(PRE_TRAINED)
+
+  return w2v
+
+def document_vector(w2v, x_train, x_dev, x_test):
+  x_train = [np.mean([w2v[word] for word in doc if word in w2v] or [np.zeros(300)], axis = 0) for doc in x_train]
+  x_dev = [np.mean([w2v[word] for word in doc if word in w2v] or [np.zeros(300)], axis = 0) for doc in x_dev]
+  x_test = [np.mean([w2v[word] for word in doc if word in w2v] or [np.zeros(300)], axis = 0) for doc in x_test]
+
+  return x_train, x_dev, x_test
+
+def basic_config():
+  INPUT_DIM = 300
+  BATCH_SIZE = 5
+
+  return INPUT_DIM, BATCH_SIZE
+
+def init_model(INPUT_DIM):
+  nn_model = NeuralNetwork(INPUT_DIM)
+  criterion = torch.nn.BCELoss()
+  optimizer = torch.optim.SGD(nn_model.parameters(), lr = 0.1)
+
+  return nn_model, optimizer, criterion
+
+def train(nn_model, BATCH_SIZE, criterion, optimizer, x_train, y_train):
+  for epoch in range(5):
+    nn_model.train()
+    for i in range(0, y_train.shape[0], BATCH_SIZE):
+        X = x_train[i:i+BATCH_SIZE]
+        X = torch.tensor(X)
+        y = y_train[i:i+BATCH_SIZE]
+        y = torch.tensor(y.astype(np.float32).to_numpy()).reshape(-1, 1)
+        outputs = nn_model(X.float())
+        loss = criterion(outputs, y)
+        optimizer.zero_grad()
+        loss.backward()
+        optimizer.step()
+
+def prediction(nn_model, BATCH_SIZE, x_dev, x_test):
+  y_dev, y_test = [], []
+  nn_model.eval()
+  with torch.no_grad():
+      for i in range(0, len(x_dev), BATCH_SIZE):
+          X = x_dev[i:i+BATCH_SIZE]
+          X = torch.tensor(X)
+          outputs = nn_model(X.float())
+          prediction = (outputs > 0.5)
+          y_dev += prediction.tolist()
+      for i in range(0, len(x_test), BATCH_SIZE):
+          X = x_test[i:i+BATCH_SIZE]
+          X = torch.tensor(X)
+          outputs = nn_model(X.float())
+          prediction = (outputs > 0.5)
+          y_test += prediction.tolist()
+
+  return y_dev, y_test
+
+def get_result(y_dev, y_test):
+  np.asarray(y_dev, dtype = np.int32).tofile(PATHS[4], sep='\n')
+  np.asarray(y_test, dtype = np.int32).tofile(PATHS[5], sep='\n')
+
+def main():
+  x_train, y_train, x_dev, x_test = get_data(FEATURES, PATHS)
+  x_train, y_train, x_dev, x_test = preprocess(x_train, y_train, x_dev, x_test)
+  x_train, x_dev, x_test = tokenize(x_train, x_dev, x_test)
+  w2v = use_word2vec()
+  x_train, x_dev, x_test = document_vector(w2v, x_train, x_dev, x_test)
+  INPUT_DIM, BATCH_SIZE = basic_config()
+  nn_model, optimizer, criterion = init_model(INPUT_DIM)
+  train(nn_model, BATCH_SIZE, criterion, optimizer, x_train, y_train)
+  y_dev, y_test = prediction(nn_model, BATCH_SIZE, x_dev, x_test)
+  get_result(y_dev, y_test)
+
+if _name_ == '_main_':
+  main()

roberta.py

@@ -0,0 +1,84 @@
+from transformers import AutoTokenizer, AutoModelForSequenceClassification, TrainingArguments, Trainer
+import torch
+
+PATHS = ['train/in.tsv', 'train/expected.tsv', 'dev-0/in.tsv', 'test-A/in.tsv', './dev-0/out.tsv', './test-A/out.tsv']
+OUTPUT_PATHS = ['dev-0/out.tsv', 'test-A/out.tsv']
+PRE_TRAINED = ['roberta-base']
+
+def get_data(path):
+    data = []
+    with open(path, encoding='utf-8') as f:
+        data = f.readlines()
+
+    return data
+
+def generate_output(path, trainer, X_data):
+    data = []
+    with open(path, encoding='utf-8') as f:
+        for result in trainer.predict(X_data):
+            f.write(str(result) + '\n')
+
+class IMDbDataset(torch.utils.data.Dataset):
+    def __init__(self, encodings, labels):
+        self.encodings = encodings
+        self.labels = labels
+
+    def __getitem__(self, idx):
+        item = {key: torch.tensor(val[idx]) for key, val in self.encodings.items()}
+        item['labels'] = torch.tensor(self.labels[idx])
+        return item
+
+    def __len__(self):
+        return len(self.labels)
+
+def prepare(data_train_X, data_train_Y):
+    tokenizer = AutoTokenizer.from_pretrained(PRE_TRAINED[0])
+    model = AutoModelForSequenceClassification.from_pretrained(PRE_TRAINED[0], num_labels=2)
+    device = torch.device("cpu")
+    model.to(device)
+    encoded_input  = tokenizer([text[0] for text in list(zip(data_train_X, data_train_Y))], truncation=True, padding=True)
+    train_dataset = IMDbDataset(encoded_input , [int(text[1]) for text in list(zip(data_train_X, data_train_Y))])
+
+    return train_dataset, model
+    
+def training(train_dataset, model):
+    training_args = TrainingArguments(
+    output_dir='./results',          # output directory
+    num_train_epochs=3,              # total number of training epochs
+    per_device_train_batch_size=16,  # batch size per device during training
+    per_device_eval_batch_size=64,   # batch size for evaluation
+    warmup_steps=500,                # number of warmup steps for learning rate scheduler
+    weight_decay=0.01,               # strength of weight decay
+    logging_dir='./logs',            # directory for storing logs
+    logging_steps=10,
+    )
+
+    trainer = Trainer(
+        model=model,                         # the instantiated Transformers model to be trained
+        args=training_args,                  # training arguments, defined above
+        train_dataset=train_dataset,         # training dataset
+    )
+
+    trainer.train()
+
+    return trainer
+
+def main():
+    #data
+    X_train = get_data(PATHS[0])
+    y_train = get_data(PATHS[1])
+    X_dev = get_data(PATHS[2])
+    X_test = get_data(PATHS[3])
+
+    #prepare
+    train_dataset, model = prepare(X_train, y_train)
+
+    #trainer
+    trainer = training(train_dataset, model)
+
+    #output
+    generate_output(OUTPUT_PATHS[0], trainer, X_dev)
+    generate_output(OUTPUT_PATHS[1], trainer, X_test)
+
+if __name__ == '__main__':
+    main()