Spaces:
Runtime error
Runtime error
Delete app.py
Browse files
app.py
DELETED
@@ -1,100 +0,0 @@
|
|
1 |
-
import gradio as gr
|
2 |
-
import re
|
3 |
-
|
4 |
-
from gradio.mix import Parallel
|
5 |
-
from transformers import (
|
6 |
-
AutoTokenizer,
|
7 |
-
AutoModelForSeq2SeqLM
|
8 |
-
)
|
9 |
-
|
10 |
-
def clean_text(text):
|
11 |
-
text = text.encode("ascii", errors="ignore").decode(
|
12 |
-
"ascii"
|
13 |
-
) # remove non-ascii, Chinese characters
|
14 |
-
text = re.sub(r"\n", " ", text)
|
15 |
-
text = re.sub(r"\n\n", " ", text)
|
16 |
-
text = re.sub(r"\t", " ", text)
|
17 |
-
text = text.strip(" ")
|
18 |
-
text = re.sub(
|
19 |
-
" +", " ", text
|
20 |
-
).strip() # get rid of multiple spaces and replace with a single
|
21 |
-
return text
|
22 |
-
|
23 |
-
|
24 |
-
modchoice_1 = "chinhon/bart-large-cnn-summarizer_03"
|
25 |
-
|
26 |
-
def summarizer1(text):
|
27 |
-
input_text = clean_text(text)
|
28 |
-
|
29 |
-
tokenizer_1 = AutoTokenizer.from_pretrained(modchoice_1)
|
30 |
-
|
31 |
-
model_1 = AutoModelForSeq2SeqLM.from_pretrained(modchoice_1)
|
32 |
-
|
33 |
-
with tokenizer_1.as_target_tokenizer():
|
34 |
-
batch = tokenizer_1(
|
35 |
-
input_text, truncation=True, padding="longest", return_tensors="pt"
|
36 |
-
)
|
37 |
-
|
38 |
-
raw_1 = model_1.generate(**batch)
|
39 |
-
|
40 |
-
summary_1 = tokenizer_1.batch_decode(raw_1, skip_special_tokens=True)
|
41 |
-
|
42 |
-
summed_1 = summary_1[0]
|
43 |
-
|
44 |
-
lines1 = summed_1.split(". ")
|
45 |
-
|
46 |
-
for i in range(len(lines1)):
|
47 |
-
lines1[i] = "* " + lines1[i]
|
48 |
-
|
49 |
-
summ_bullet1 = "\n".join(lines1)
|
50 |
-
|
51 |
-
return summ_bullet1
|
52 |
-
|
53 |
-
|
54 |
-
summary1 = gr.Interface(
|
55 |
-
fn=summarizer1, inputs=gr.inputs.Textbox(), outputs=gr.outputs.Textbox(label="")
|
56 |
-
)
|
57 |
-
|
58 |
-
|
59 |
-
modchoice_2 = (
|
60 |
-
"chinhon/pegasus-newsroom-summarizer_02"
|
61 |
-
)
|
62 |
-
|
63 |
-
def summarizer2(text):
|
64 |
-
input_text = clean_text(text)
|
65 |
-
|
66 |
-
tokenizer_2 = AutoTokenizer.from_pretrained(modchoice_2)
|
67 |
-
|
68 |
-
model_2 = AutoModelForSeq2SeqLM.from_pretrained(modchoice_2)
|
69 |
-
|
70 |
-
with tokenizer_2.as_target_tokenizer():
|
71 |
-
batch = tokenizer_2(
|
72 |
-
input_text, truncation=True, padding="longest", return_tensors="pt"
|
73 |
-
)
|
74 |
-
|
75 |
-
raw_2 = model_2.generate(**batch)
|
76 |
-
|
77 |
-
summary_2 = tokenizer_2.batch_decode(raw_2, skip_special_tokens=True)
|
78 |
-
|
79 |
-
summed_2 = summary_2[0]
|
80 |
-
|
81 |
-
lines2 = summed_2.split(". ")
|
82 |
-
|
83 |
-
for i in range(len(lines2)):
|
84 |
-
lines2[i] = "* " + lines2[i]
|
85 |
-
|
86 |
-
summ_bullet2 = "\n".join(lines2)
|
87 |
-
|
88 |
-
return summ_bullet2
|
89 |
-
|
90 |
-
summary2 = gr.Interface(
|
91 |
-
fn=summarizer2, inputs=gr.inputs.Textbox(), outputs=gr.outputs.Textbox(label="")
|
92 |
-
)
|
93 |
-
|
94 |
-
|
95 |
-
Parallel(
|
96 |
-
summary1,
|
97 |
-
summary2,
|
98 |
-
title="Compare 2 AI Summarizers",
|
99 |
-
inputs=gr.inputs.Textbox(lines=20, label="Paste your news story here, and choose from 2 suggested summaries"),
|
100 |
-
).launch(enable_queue=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|