Spaces:
Runtime error
Runtime error
Added genre guessing
Browse files- pages/genre_model.py +66 -0
pages/genre_model.py
ADDED
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import torch
|
2 |
+
from torch import nn
|
3 |
+
from joblib import load
|
4 |
+
import textwrap
|
5 |
+
import streamlit as st
|
6 |
+
|
7 |
+
device = 'cpu'
|
8 |
+
|
9 |
+
|
10 |
+
|
11 |
+
class GenreNet(nn.Module):
|
12 |
+
def __init__(self, config):
|
13 |
+
super().__init__()
|
14 |
+
# параметры сетиnspose arrayroupout']
|
15 |
+
self.dropout = config['dropout']
|
16 |
+
self.out_range = config['out_range']
|
17 |
+
|
18 |
+
|
19 |
+
# финальный полносвязный слой для пронгоза оценки
|
20 |
+
self.head = nn.Sequential(
|
21 |
+
nn.Linear(312, 256),
|
22 |
+
nn.Dropout(self.dropout[0]),
|
23 |
+
nn.ReLU(),
|
24 |
+
nn.Linear(256, 128),
|
25 |
+
nn.Dropout(self.dropout[0]),
|
26 |
+
nn.ReLU(),
|
27 |
+
nn.Linear(128, 64),
|
28 |
+
nn.Dropout(self.dropout[0]),
|
29 |
+
nn.ReLU(),
|
30 |
+
nn.Linear(64, 1),
|
31 |
+
)
|
32 |
+
|
33 |
+
|
34 |
+
|
35 |
+
def forward(self, emb):
|
36 |
+
x = torch.sigmoid(self.head(emb))
|
37 |
+
x = x * (self.out_range[1] - self.out_range[0]) + self.out_range[0]
|
38 |
+
return(x)
|
39 |
+
|
40 |
+
config = {
|
41 |
+
'dropout': [.5],
|
42 |
+
'out_range': [1.,5.] # для номировки выходных оценок
|
43 |
+
}
|
44 |
+
|
45 |
+
bert = load('./model.joblib')
|
46 |
+
model = GenreNet(config)
|
47 |
+
model.load_state_dict(torch.load('./pages/weights_los065_ep100_lr0001_lay256_128_64_1.pt', map_location=device))
|
48 |
+
tokenizer = load('./tokenizer.joblib')
|
49 |
+
|
50 |
+
def embed_bert_cls(text, model, tokenizer):
|
51 |
+
t = tokenizer(text, padding=True, truncation=True, return_tensors='pt')
|
52 |
+
with torch.no_grad():
|
53 |
+
model_output = model(**{k: v.to(device) for k, v in t.items()})
|
54 |
+
embeddings = model_output.last_hidden_state[:, 0, :]
|
55 |
+
embeddings = torch.nn.functional.normalize(embeddings)
|
56 |
+
return embeddings[0]
|
57 |
+
|
58 |
+
genre = {1 : 'Романтика', 2:'Поэзия', 3:'Детектив', 4:'Приключения', 5:'Фантастика', }
|
59 |
+
|
60 |
+
prompt = st.text_input('Узнаем жанр!',)
|
61 |
+
if len(prompt) > 1:
|
62 |
+
with torch.inference_mode():
|
63 |
+
prompt_embedding = embed_bert_cls([prompt], bert, tokenizer)
|
64 |
+
out = model(prompt_embedding).cpu().numpy()
|
65 |
+
#for out_ in out:
|
66 |
+
st.write('Предполагаемый жанр:', genre[int(round(out.item(), 0))])
|