nami
nami is python package
Installation
pip install --upgrade nami
Features (nami-1.2.0.1)
get_dataset
from nami.datasets.ImageNet import get_dataset
dataset = get_dataset(noun='STR', dimension=(INT,INT), max=INT, timeout=FLOAT, save_to='STR')get 'INT*INT dimenstion' of 'noun' image dataset from ImageNet.
timeout - [from 0.1 - 1.0] maximum time request for each image URL. max - number of images dataset. save_to - save the dataset by '.npy' format.
load_dataset (KME)
from nami.datasets.kme import load_data
(X_train, y_train), (X_test, y_test) = load_data(is_split=True. test_size=0.2)
(X, y) = load_data(is_split=False)Tokenizer class
from nami.AI.kme_tokenize import Tokenizer
tokenizer = Tokenizer()
text_arr = ['methyl methanoate', 'ethane', '(hydroxymethylamino)oxy-methoxymethanol']
```python
tokenizer.fit_on_text(text_arr)fit_on_text(sentences=)
sentences: take array of string to make bag of words (word2index & index2word)
train_seq = tokenizer.text_to_sequences(text_arr, method_pad='pre')text_to_sequences(sentences= , method_pad='post')
sentences: take array of string to preprocessing text to numeric
method_pad: ('pre', 'post') make zero padding
train_seq
[[ 0 0 0 0 0 0 0 4 5 6 4 7 8]
[ 0 0 0 0 0 0 0 0 0 0 0 9 10]
[11 12 4 5 13 14 15 16 4 15 4 7 17]]
test_arr = ['2-(4-methoxyphenyl)-2-oxoacetic acid']
test_seq = tokenizer.text_to_sequences(test_arr)
# [[11, 14, 18, 13, 14, 4, 22, 3, 5, 21, 14, 11, 14, 3, 3, 3]]
test_text = tokenizer.sequences_to_text(test_seq)
# [['2', '-', '(', '4', '-', 'meth', 'oxy', '<unk>', 'yl', ')', '-', '2', '-', '<unk>', '<unk>', '<unk>']]
