Spaces:
Sleeping
Sleeping
EbubeJohnEnyi
commited on
Commit
•
c540671
1
Parent(s):
79ede05
Update app.py
Browse files
app.py
CHANGED
@@ -1,15 +1,86 @@
|
|
1 |
-
import streamlit as st
|
2 |
-
from transformers import pipeline
|
3 |
-
from flask import Flask, render_template, request
|
4 |
from transformers import GPT2LMHeadModel, GPT2Tokenizer
|
5 |
from sklearn.feature_extraction.text import CountVectorizer
|
6 |
from sklearn.metrics.pairwise import cosine_similarity
|
7 |
import json
|
8 |
|
|
|
|
|
9 |
|
10 |
-
|
11 |
-
|
12 |
|
13 |
-
|
14 |
-
|
15 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
from transformers import GPT2LMHeadModel, GPT2Tokenizer
|
2 |
from sklearn.feature_extraction.text import CountVectorizer
|
3 |
from sklearn.metrics.pairwise import cosine_similarity
|
4 |
import json
|
5 |
|
6 |
+
# Set the path to your dataset file
|
7 |
+
dataset_path = 'path/to/your/dataset.json'
|
8 |
|
9 |
+
tokenizer = GPT2Tokenizer.from_pretrained('gpt2')
|
10 |
+
model = GPT2LMHeadModel.from_pretrained('gpt2')
|
11 |
|
12 |
+
def compare_sentences(sentence1, sentence2):
|
13 |
+
vectorizer = CountVectorizer().fit_transform([sentence1, sentence2])
|
14 |
+
similarity = cosine_similarity(vectorizer)
|
15 |
+
similarity_score = similarity[0, 1]
|
16 |
+
return similarity_score
|
17 |
+
|
18 |
+
def generate_gpt2_response(question):
|
19 |
+
input_ids = tokenizer.encode(question, return_tensors='pt')
|
20 |
+
generated_output = model.generate(input_ids, max_length=len(input_ids[0]) + 100,
|
21 |
+
num_beams=5,
|
22 |
+
no_repeat_ngram_size=2,
|
23 |
+
top_k=10,
|
24 |
+
top_p=1,
|
25 |
+
temperature=0.9)
|
26 |
+
generated_response = tokenizer.decode(generated_output[0], skip_special_tokens=True)
|
27 |
+
return generated_response
|
28 |
+
|
29 |
+
def find_question_and_answer(dataset_file, question):
|
30 |
+
with open(dataset_file, "r") as json_file:
|
31 |
+
data = json.load(json_file)
|
32 |
+
|
33 |
+
question = question.lower()
|
34 |
+
|
35 |
+
max_similarity = 0
|
36 |
+
selected_response = None
|
37 |
+
|
38 |
+
for q_and_a in data["questions"]:
|
39 |
+
response_message = q_and_a["response"].lower()
|
40 |
+
similarity_score = compare_sentences(question, response_message)
|
41 |
+
|
42 |
+
if similarity_score > max_similarity:
|
43 |
+
max_similarity = similarity_score
|
44 |
+
selected_response = q_and_a["response"]
|
45 |
+
|
46 |
+
# Set a threshold for similarity score to switch to GPT-2
|
47 |
+
similarity_threshold = 0.4 # Adjust this threshold as needed
|
48 |
+
if max_similarity < similarity_threshold:
|
49 |
+
generated_response = generate_gpt2_response(question)
|
50 |
+
selected_response = generated_response
|
51 |
+
|
52 |
+
# Fallback to a default message if no suitable response is found
|
53 |
+
if selected_response is None:
|
54 |
+
selected_response = "CHAT BOT --> I'm sorry, I don't have data about that.\n"
|
55 |
+
|
56 |
+
return selected_response
|
57 |
+
|
58 |
+
# Example usage
|
59 |
+
user_input = input("Ask a question: ")
|
60 |
+
response = find_question_and_answer(dataset_path, user_input)
|
61 |
+
print(response)
|
62 |
+
|
63 |
+
|
64 |
+
|
65 |
+
|
66 |
+
|
67 |
+
|
68 |
+
|
69 |
+
|
70 |
+
|
71 |
+
|
72 |
+
# import streamlit as st
|
73 |
+
# from transformers import pipeline
|
74 |
+
# from flask import Flask, render_template, request
|
75 |
+
# from transformers import GPT2LMHeadModel, GPT2Tokenizer
|
76 |
+
# from sklearn.feature_extraction.text import CountVectorizer
|
77 |
+
# from sklearn.metrics.pairwise import cosine_similarity
|
78 |
+
# import json
|
79 |
+
|
80 |
+
|
81 |
+
# pipe = pipeline('sentiment-analysis')
|
82 |
+
# text = st.text_area('Enter your text here: ')
|
83 |
+
|
84 |
+
# if text:
|
85 |
+
# out = pipe(text)
|
86 |
+
# print(out)
|