215 lines
7.7 KiB
Plaintext
215 lines
7.7 KiB
Plaintext
{
|
|
"cells": [
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 2,
|
|
"id": "308eb052",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"import torch\n",
|
|
"import gensim.downloader as downloader\n",
|
|
"import pandas as pd\n",
|
|
"import csv\n",
|
|
"from nltk.tokenize import word_tokenize as tokenize\n",
|
|
"import numpy as np"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 7,
|
|
"id": "212c0f2c",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"class NeuralNetworkModel(torch.nn.Module):\n",
|
|
" \n",
|
|
" def __init__(self, input_size, hidden_size, num_classes):\n",
|
|
" super(NeuralNetworkModel, self).__init__()\n",
|
|
" self.fc1 = torch.nn.Linear(input_size,hidden_size)\n",
|
|
" self.fc2 = torch.nn.Linear(hidden_size,num_classes)\n",
|
|
" \n",
|
|
" def forward(self,x):\n",
|
|
" x = self.fc1(x)\n",
|
|
" x = torch.relu(x)\n",
|
|
" x = self.fc2(x)\n",
|
|
" x = torch.sigmoid(x)\n",
|
|
" return x"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 4,
|
|
"id": "afd13d8c",
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"[==================================================] 100.0% 1662.8/1662.8MB downloaded\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"w2v = downloader.load('word2vec-google-news-300')"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 9,
|
|
"id": "f68b0a7c",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"#model + settings\n",
|
|
"\n",
|
|
"nn = NeuralNetworkModel(300,300,1)\n",
|
|
"crit = torch.nn.BCELoss()\n",
|
|
"opti = torch.optim.SGD(nn.parameters(), lr=0.08)\n",
|
|
"BATCH_SIZE = 5\n",
|
|
"epochs = 5"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 12,
|
|
"id": "f0713eab",
|
|
"metadata": {
|
|
"scrolled": true
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"#trening\n",
|
|
"\n",
|
|
"#wczytanie danych\n",
|
|
"train_data_in = pd.read_csv('train/in.tsv.xz', compression='xz', header=None, error_bad_lines=False, quoting=csv.QUOTE_NONE, sep='\\t', nrows=3000)\n",
|
|
"train_data_ex = pd.read_csv('train/expected.tsv', header=None, error_bad_lines=False, quoting=csv.QUOTE_NONE, sep='\\t', nrows=3000)\n",
|
|
"\n",
|
|
"#preprocessing\n",
|
|
"train_in = train_data_in[0].str.lower()\n",
|
|
"train_in = [tokenize(line) for line in train_in]\n",
|
|
"train_in = [np.mean([w2v[x] for x in data if x in w2v] or [np.zeros(300)], axis=0) for data in train_in]\n",
|
|
"train_ex = train_data_ex[0]\n",
|
|
"\n",
|
|
"for epoch in range(epochs):\n",
|
|
" nn.train()\n",
|
|
" for i in range(0,train_data_ex.shape[0],BATCH_SIZE):\n",
|
|
" x = train_in[i:i + BATCH_SIZE]\n",
|
|
" x = torch.tensor(x)\n",
|
|
" y = train_ex[i:i + BATCH_SIZE]\n",
|
|
" y = torch.tensor(y.astype(np.float32).to_numpy()).reshape(-1,1)\n",
|
|
" \n",
|
|
" opti.zero_grad()\n",
|
|
" y_pred = nn(x.float())\n",
|
|
" loss = crit(y_pred,y)\n",
|
|
" loss.backward()\n",
|
|
" opti.step()"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 27,
|
|
"id": "1ab1dce0",
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"name": "stderr",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"<ipython-input-27-b22b834acfd4>:21: FutureWarning: The input object of type 'Tensor' is an array-like implementing one of the corresponding protocols (`__array__`, `__array_interface__` or `__array_struct__`); but not a sequence (or 0-D). In the future, this object will be coerced as if it was first converted using `np.array(obj)`. To retain the old behaviour, you have to either modify the type 'Tensor', or assign to an empty array created with `np.empty(correct_shape, dtype=object)`.\n",
|
|
" np.asarray(dev0_y, dtype=np.int32).tofile('dev-0/out.tsv', sep='\\n')\n",
|
|
"<ipython-input-27-b22b834acfd4>:21: DeprecationWarning: setting an array element with a sequence. This was supported in some cases where the elements are arrays with a single element. For example `np.array([1, np.array([2])], dtype=int)`. In the future this will raise the same ValueError as `np.array([1, [2]], dtype=int)`.\n",
|
|
" np.asarray(dev0_y, dtype=np.int32).tofile('dev-0/out.tsv', sep='\\n')\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"#dev-0 predict\n",
|
|
"\n",
|
|
"#wczytanie danych\n",
|
|
"dev0_data = pd.read_csv('dev-0/in.tsv.xz', compression='xz', header=None, error_bad_lines=False, quoting=csv.QUOTE_NONE, sep='\\t')\n",
|
|
"dev0_data = dev0_data[0].str.lower()\n",
|
|
"dev0_data = [tokenize(line) for line in dev0_data]\n",
|
|
"dev0_data = [np.mean([w2v[x] for x in data if x in w2v] or [np.zeros(300)], axis=0) for data in dev0_data]\n",
|
|
"\n",
|
|
"dev0_y=[]\n",
|
|
"nn.eval()\n",
|
|
"with torch.no_grad():\n",
|
|
" for i in range(0, len(dev0_data), BATCH_SIZE):\n",
|
|
" x = dev0_data[i:i + BATCH_SIZE]\n",
|
|
" x = torch.tensor(x)\n",
|
|
" dev0_y_pred = nn(x.float())\n",
|
|
"\n",
|
|
" dev0_y_prediction = (dev0_y_pred > 0.5)\n",
|
|
" dev0_y.extend(dev0_y_prediction)\n",
|
|
" \n",
|
|
"#zapis wyników\n",
|
|
"np.asarray(dev0_y, dtype=np.int32).tofile('dev-0/out.tsv', sep='\\n')"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 28,
|
|
"id": "22941828",
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"name": "stderr",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"<ipython-input-28-ed4376367760>:21: FutureWarning: The input object of type 'Tensor' is an array-like implementing one of the corresponding protocols (`__array__`, `__array_interface__` or `__array_struct__`); but not a sequence (or 0-D). In the future, this object will be coerced as if it was first converted using `np.array(obj)`. To retain the old behaviour, you have to either modify the type 'Tensor', or assign to an empty array created with `np.empty(correct_shape, dtype=object)`.\n",
|
|
" np.asarray(testA_y, dtype=np.int32).tofile('test-A/out.tsv', sep='\\n')\n",
|
|
"<ipython-input-28-ed4376367760>:21: DeprecationWarning: setting an array element with a sequence. This was supported in some cases where the elements are arrays with a single element. For example `np.array([1, np.array([2])], dtype=int)`. In the future this will raise the same ValueError as `np.array([1, [2]], dtype=int)`.\n",
|
|
" np.asarray(testA_y, dtype=np.int32).tofile('test-A/out.tsv', sep='\\n')\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"#test-A predict\n",
|
|
"\n",
|
|
"#wczytanie danych\n",
|
|
"testA_data = pd.read_csv('test-A/in.tsv.xz', compression='xz', header=None, quoting=csv.QUOTE_NONE, sep='\\t')\n",
|
|
"testA_data = testA_data[0].str.lower()\n",
|
|
"testA_data = [tokenize(line) for line in testA_data]\n",
|
|
"testA_data = [np.mean([w2v[x] for x in data if x in w2v] or [np.zeros(300)], axis=0) for data in testA_data]\n",
|
|
"\n",
|
|
"testA_y=[]\n",
|
|
"nn.eval()\n",
|
|
"with torch.no_grad():\n",
|
|
" for i in range(0, len(testA_data), BATCH_SIZE):\n",
|
|
" x = testA_data[i:i + BATCH_SIZE]\n",
|
|
" x = torch.tensor(x)\n",
|
|
" testA_y_pred = nn(x.float())\n",
|
|
"\n",
|
|
" testA_y_prediction = (testA_y_pred > 0.5)\n",
|
|
" testA_y.extend(testA_y_prediction)\n",
|
|
" \n",
|
|
"#zapis wyników\n",
|
|
"np.asarray(testA_y, dtype=np.int32).tofile('test-A/out.tsv', sep='\\n')"
|
|
]
|
|
}
|
|
],
|
|
"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": "3.8.8"
|
|
}
|
|
},
|
|
"nbformat": 4,
|
|
"nbformat_minor": 5
|
|
}
|