Delete app.py
Browse files
app.py
DELETED
@@ -1,128 +0,0 @@
|
|
1 |
-
import json
|
2 |
-
import string
|
3 |
-
import random
|
4 |
-
import nltk
|
5 |
-
import numpy as np
|
6 |
-
import tensorflow as tf
|
7 |
-
from nltk.stem import WordNetLemmatizer
|
8 |
-
from tensorflow.keras import Sequential
|
9 |
-
from tensorflow.keras.layers import Dense, Dropout
|
10 |
-
from tensorflow.keras.optimizers.schedules import ExponentialDecay
|
11 |
-
import gradio as gr
|
12 |
-
|
13 |
-
# Download NLTK resources
|
14 |
-
nltk.download("punkt")
|
15 |
-
nltk.download("wordnet")
|
16 |
-
|
17 |
-
# Load the intents JSON file
|
18 |
-
with open('intents.json', 'r') as file:
|
19 |
-
data = json.load(file)
|
20 |
-
|
21 |
-
# Preprocess and prepare data
|
22 |
-
lemmatizer = WordNetLemmatizer()
|
23 |
-
words = []
|
24 |
-
classes = []
|
25 |
-
data_X = []
|
26 |
-
data_Y = []
|
27 |
-
|
28 |
-
for intent in data["intents"]:
|
29 |
-
for pattern in intent["patterns"]:
|
30 |
-
tokens = nltk.word_tokenize(pattern)
|
31 |
-
words.extend(tokens)
|
32 |
-
data_X.append(pattern)
|
33 |
-
data_Y.append(intent["tag"])
|
34 |
-
if intent["tag"] not in classes:
|
35 |
-
classes.append(intent["tag"])
|
36 |
-
|
37 |
-
words = [lemmatizer.lemmatize(word.lower()) for word in words if word not in string.punctuation]
|
38 |
-
words = sorted(set(words))
|
39 |
-
classes = sorted(set(classes))
|
40 |
-
|
41 |
-
# Bag of Words model
|
42 |
-
training = []
|
43 |
-
out_empty = [0] * len(classes)
|
44 |
-
|
45 |
-
for idx, doc in enumerate(data_X):
|
46 |
-
bow = []
|
47 |
-
text = lemmatizer.lemmatize(doc.lower())
|
48 |
-
for word in words:
|
49 |
-
bow.append(1) if word in text else bow.append(0)
|
50 |
-
output_row = list(out_empty)
|
51 |
-
output_row[classes.index(data_Y[idx])] = 1
|
52 |
-
training.append([bow, output_row])
|
53 |
-
|
54 |
-
random.shuffle(training)
|
55 |
-
training = np.array(training, dtype=object)
|
56 |
-
train_X = np.array(list(training[:, 0]), dtype=np.float32)
|
57 |
-
train_Y = np.array(list(training[:, 1]), dtype=np.float32)
|
58 |
-
|
59 |
-
# Define learning rate schedule
|
60 |
-
initial_learning_rate = 0.01
|
61 |
-
lr_schedule = ExponentialDecay(
|
62 |
-
initial_learning_rate=initial_learning_rate,
|
63 |
-
decay_steps=1000,
|
64 |
-
decay_rate=0.9,
|
65 |
-
staircase=True
|
66 |
-
)
|
67 |
-
|
68 |
-
# Build the neural network model
|
69 |
-
model = Sequential([
|
70 |
-
Dense(128, input_shape=(len(train_X[0]),), activation="relu"),
|
71 |
-
Dropout(0.5),
|
72 |
-
Dense(64, activation="relu"),
|
73 |
-
Dropout(0.5),
|
74 |
-
Dense(len(train_Y[0]), activation="softmax")
|
75 |
-
])
|
76 |
-
|
77 |
-
adam = tf.keras.optimizers.Adam(learning_rate=lr_schedule)
|
78 |
-
model.compile(loss='categorical_crossentropy', optimizer=adam, metrics=["accuracy"])
|
79 |
-
model.fit(x=train_X, y=train_Y, epochs=150, verbose=1)
|
80 |
-
|
81 |
-
# Define the helper functions
|
82 |
-
def clean_text(text):
|
83 |
-
tokens = nltk.word_tokenize(text)
|
84 |
-
tokens = (lemmatizer.lemmatize(word) for word in tokens)
|
85 |
-
return tokens
|
86 |
-
|
87 |
-
def bag_of_words(text, vocab):
|
88 |
-
tokens = clean_text(text)
|
89 |
-
bow = [0] * len(vocab)
|
90 |
-
for w in tokens:
|
91 |
-
if w in vocab:
|
92 |
-
bow[vocab.index(w)] = 1
|
93 |
-
return np.array(bow, dtype=np.float32)
|
94 |
-
|
95 |
-
def pred_class(text, vocab, labels):
|
96 |
-
bow = bag_of_words(text, vocab)
|
97 |
-
result = model.predict(np.array([bow]))[0]
|
98 |
-
thresh = 0.5
|
99 |
-
y_pred = [[indx, res] for indx, res in enumerate(result) if res > thresh]
|
100 |
-
y_pred.sort(key=lambda x: x[1], reverse=True)
|
101 |
-
return_list = [labels[r[0]] for r in y_pred]
|
102 |
-
return return_list
|
103 |
-
|
104 |
-
def get_response(intents_list, intents_json):
|
105 |
-
if len(intents_list) == 0:
|
106 |
-
return "Sorry! I didn't understand."
|
107 |
-
else:
|
108 |
-
tag = intents_list[0]
|
109 |
-
list_of_intents = intents_json["intents"]
|
110 |
-
for i in list_of_intents:
|
111 |
-
if i["tag"] == tag:
|
112 |
-
return random.choice(i["responses"])
|
113 |
-
|
114 |
-
def chatbot_response(message):
|
115 |
-
intents = pred_class(message, words, classes)
|
116 |
-
result = get_response(intents, data)
|
117 |
-
return result
|
118 |
-
|
119 |
-
# Create a Gradio interface
|
120 |
-
interface = gr.Interface(
|
121 |
-
fn=chatbot_response,
|
122 |
-
inputs=gr.Textbox(label="You:"),
|
123 |
-
outputs=gr.Textbox(label="ChatBot:"),
|
124 |
-
title="ChatBot",
|
125 |
-
description="Chat with the chatbot."
|
126 |
-
)
|
127 |
-
|
128 |
-
interface.launch()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|