-
Notifications
You must be signed in to change notification settings - Fork 1
/
app.py
59 lines (50 loc) · 2.12 KB
/
app.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
import os
import streamlit as st
import torch
import numpy as np
import pandas as pd
from collections import OrderedDict
from network import CharacterCNN, CharacterCNNClassifier, CharacterCNNEmbedding
from utils import predict_some_block
parent = os.path.dirname(os.path.abspath(__file__))
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
state_dict_emb = torch.load(os.path.join(parent, 'pretrain', 'model_fold1.pth'), map_location=device)
state_dict_emb = OrderedDict(list(state_dict_emb.items())[0:1])
embed = CharacterCNNEmbedding().to(device)
embed.load_state_dict(state_dict_emb)
classifier = dict()
for artist in os.listdir(os.path.join(parent, 'models')):
model_list = []
for file in os.listdir(os.path.join(parent, 'models', artist)):
state_dict = torch.load(os.path.join(parent, 'models', artist, file), map_location=device)
clf = CharacterCNNClassifier(2).to(device)
clf.load_state_dict(state_dict)
model_list.append(clf)
classifier[artist] = model_list
artists = list(classifier.keys())
st.title('その歌詞はどのアーティストっぽい?')
# TEXT BOX
text = st.text_area('歌詞を入力してください。歌詞全体を入力するときは、ブロックごとに1行空けるようにしてください。', height=200)
text_list = text.split('\n\n')
# PREDICT
if text == '':
prob = np.zeros(len(artists))
st.error('歌詞を入力してください!')
else:
with st.spinner('Inference...'):
prob = predict_some_block(text_list, embed, classifier, device)
prob = prob.mean(axis=0)
st.success(f'その歌詞は「{artists[np.argmax(prob)]}」っぽい!')
# RESULT
df = pd.DataFrame({'artist':artists,'probability':prob*100})
df.sort_values('probability', ascending=False, inplace=True)
df.reset_index(drop=True, inplace=True)
df.set_index('artist', inplace=True)
if text != '':
# TOP3
cols = st.columns(3)
for i, col in enumerate(cols):
col.metric(f'{i+1}位', df.index[i], f'{df.probability[i]:.1f}%')
# BAR CHART
st.bar_chart(df)
st.write(df)