diff --git a/python/neural_bag_of_words.py b/python/neural_bag_of_words.py
index e69de29..2744d24 100644
--- a/python/neural_bag_of_words.py
+++ b/python/neural_bag_of_words.py
@@ -0,0 +1,38 @@
+import collections
+
+import datasets
+import matplotlib.pyplot as plt
+import numpy as np
+import torch
+import torch.nn as nn
+import torch.optim as optim
+import nltk
+nltk.download('punkt')
+nltk.download('averaged_perceptron_tagger')
+import tqdm
+
+seed = 1234
+validate_size = 0.25
+
+np.random.seed(seed)
+torch.manual_seed(seed)
+torch.cuda.manual_seed(seed)
+torch.backends.cudnn.deterministic = True
+
+train_data, test_data = datasets.load_dataset("imdb", split=["train", "test"])
+
+print(nltk.pos_tag(nltk.word_tokenize("Hello there you stupid fucking whore mr parker")))
+
+def tokenize(input):
+    return {"tokens": nltk.word_tokenize(input["text"])}
+
+train_data = train_data.map(tokenize)
+test_data = test_data.map(tokenize)
+
+train_valid_data = train_data.train_test_split(test_size=validate_size)
+train_data = train_valid_data["train"]
+valid_data = train_valid_data["test"]
+
+print(train_data)
+print(test_data)
+print(valid_data)
\ No newline at end of file