Spaces:
Runtime error
Runtime error
Commit
β’
a432184
1
Parent(s):
66700ad
update - more checking user input
Browse files
app.py
CHANGED
@@ -8,25 +8,27 @@ from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
|
8 |
st.markdown("# Hello, friend!")
|
9 |
st.markdown(" This magic application going to help you with understanding of science paper topic! Cool? Yeah! ")
|
10 |
|
11 |
-
|
12 |
-
model_name_global = "allenai/scibert_scivocab_uncased"
|
13 |
-
tokenizer_ = AutoTokenizer.from_pretrained(model_name_global)
|
14 |
-
with open('./models/scibert/decode_dict.pkl', 'rb') as f:
|
15 |
-
|
|
|
|
|
16 |
|
17 |
with st.form(key="my_form"):
|
18 |
st.markdown("### π Do you want a little magic? ")
|
19 |
st.markdown(" Write your article title and abstract to textboxes bellow and I'll gues topic of your paper! ")
|
20 |
-
ce, c2, c3 = st.columns([0.07,
|
21 |
|
22 |
with c2:
|
23 |
doc_title = st.text_area(
|
24 |
-
"Paste your abstract title below (
|
25 |
height=210,
|
26 |
)
|
27 |
|
28 |
doc_abstract = st.text_area(
|
29 |
-
"Paste your abstract text below (
|
30 |
height=410,
|
31 |
)
|
32 |
|
@@ -35,6 +37,7 @@ with st.form(key="my_form"):
|
|
35 |
|
36 |
len_title = len(re.findall(r"\w+", doc_title))
|
37 |
len_abstract = len(re.findall(r"\w+", doc_abstract))
|
|
|
38 |
if len_title > MAX_WORDS_TITLE:
|
39 |
st.warning(
|
40 |
"β οΈ Your title contains "
|
@@ -50,7 +53,7 @@ with st.form(key="my_form"):
|
|
50 |
"β οΈ Your abstract contains "
|
51 |
+ str(len_abstract)
|
52 |
+ " words."
|
53 |
-
+ " Only the first
|
54 |
)
|
55 |
|
56 |
doc_abstract = doc_abstract[:MAX_WORDS_ABSTRACT]
|
@@ -60,6 +63,14 @@ with st.form(key="my_form"):
|
|
60 |
if not submit_button:
|
61 |
st.stop()
|
62 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
63 |
|
64 |
# allow_output_mutation=True
|
65 |
@st.cache(suppress_st_warning=True)
|
@@ -69,8 +80,6 @@ def load_model():
|
|
69 |
|
70 |
|
71 |
def make_predict(tokens, decode_dict):
|
72 |
-
# tokenizer_ = AutoTokenizer.from_pretrained(model_name_global)
|
73 |
-
# tokens = tokenizer_(title + abstract, return_tensors="pt")
|
74 |
|
75 |
model_ = load_model()
|
76 |
outs = model_(tokens.input_ids)
|
@@ -87,7 +96,10 @@ model_local = "models/scibert/"
|
|
87 |
|
88 |
title = doc_title
|
89 |
abstract = doc_abstract
|
90 |
-
|
|
|
|
|
|
|
91 |
|
92 |
predicts = make_predict(tokens, decode_dict)
|
93 |
|
|
|
8 |
st.markdown("# Hello, friend!")
|
9 |
st.markdown(" This magic application going to help you with understanding of science paper topic! Cool? Yeah! ")
|
10 |
|
11 |
+
try:
|
12 |
+
model_name_global = "allenai/scibert_scivocab_uncased"
|
13 |
+
tokenizer_ = AutoTokenizer.from_pretrained(model_name_global)
|
14 |
+
with open('./models/scibert/decode_dict.pkl', 'rb') as f:
|
15 |
+
decode_dict = pickle.load(f)
|
16 |
+
except ValueError:
|
17 |
+
st.error("Load tokenizer or decode answer dict goes wrong! Pls contact author alekseystepin13@gmail.com")
|
18 |
|
19 |
with st.form(key="my_form"):
|
20 |
st.markdown("### π Do you want a little magic? ")
|
21 |
st.markdown(" Write your article title and abstract to textboxes bellow and I'll gues topic of your paper! ")
|
22 |
+
ce, c2, c3 = st.columns([0.07, 7, 0.07])
|
23 |
|
24 |
with c2:
|
25 |
doc_title = st.text_area(
|
26 |
+
"Paste your abstract title below (1 to 50 words)",
|
27 |
height=210,
|
28 |
)
|
29 |
|
30 |
doc_abstract = st.text_area(
|
31 |
+
"Paste your abstract text below (1 to 500 words)",
|
32 |
height=410,
|
33 |
)
|
34 |
|
|
|
37 |
|
38 |
len_title = len(re.findall(r"\w+", doc_title))
|
39 |
len_abstract = len(re.findall(r"\w+", doc_abstract))
|
40 |
+
|
41 |
if len_title > MAX_WORDS_TITLE:
|
42 |
st.warning(
|
43 |
"β οΈ Your title contains "
|
|
|
53 |
"β οΈ Your abstract contains "
|
54 |
+ str(len_abstract)
|
55 |
+ " words."
|
56 |
+
+ " Only the first 500 words will be reviewed. Stay tuned as increased allowance is coming! π"
|
57 |
)
|
58 |
|
59 |
doc_abstract = doc_abstract[:MAX_WORDS_ABSTRACT]
|
|
|
63 |
if not submit_button:
|
64 |
st.stop()
|
65 |
|
66 |
+
if len_title < 1:
|
67 |
+
st.error("Article without any words in title? Pls give me correct title!")
|
68 |
+
st.stop()
|
69 |
+
|
70 |
+
if len_abstract < 1:
|
71 |
+
st.error("Article without any words in abstract? Pls give me correct abstract!")
|
72 |
+
st.stop()
|
73 |
+
|
74 |
|
75 |
# allow_output_mutation=True
|
76 |
@st.cache(suppress_st_warning=True)
|
|
|
80 |
|
81 |
|
82 |
def make_predict(tokens, decode_dict):
|
|
|
|
|
83 |
|
84 |
model_ = load_model()
|
85 |
outs = model_(tokens.input_ids)
|
|
|
96 |
|
97 |
title = doc_title
|
98 |
abstract = doc_abstract
|
99 |
+
try:
|
100 |
+
tokens = tokenizer_(title + abstract, return_tensors="pt")
|
101 |
+
except ValueError:
|
102 |
+
st.error("Word parsing into tokens went wrong! Is input valid? If yes, pls contact author alekseystepin13@gmail.com")
|
103 |
|
104 |
predicts = make_predict(tokens, decode_dict)
|
105 |
|