Spaces:
Build error
Build error
hellopahe commited on
Commit ·
e24946b
1
Parent(s): 77129d5
fix
Browse files
app.py
CHANGED
|
@@ -1,3 +1,5 @@
|
|
|
|
|
|
|
|
| 1 |
import numpy
|
| 2 |
import torch
|
| 3 |
import gradio as gr
|
|
@@ -64,7 +66,8 @@ class LexRank(object):
|
|
| 64 |
def __init__(self):
|
| 65 |
self.model = SentenceTransformer('paraphrase-multilingual-mpnet-base-v2')
|
| 66 |
self.ht = HarvestText()
|
| 67 |
-
|
|
|
|
| 68 |
sentences = self.ht.cut_sentences(content)
|
| 69 |
embeddings = self.model.encode(sentences, convert_to_tensor=True).cpu()
|
| 70 |
|
|
@@ -77,7 +80,7 @@ class LexRank(object):
|
|
| 77 |
# We argsort so that the first element is the sentence with the highest score
|
| 78 |
most_central_sentence_indices = numpy.argsort(-centrality_scores)
|
| 79 |
|
| 80 |
-
num = 100
|
| 81 |
res = []
|
| 82 |
for index in most_central_sentence_indices:
|
| 83 |
if num < 0:
|
|
@@ -96,7 +99,8 @@ lex = LexRank()
|
|
| 96 |
|
| 97 |
|
| 98 |
def randeng_extract(content):
|
| 99 |
-
|
|
|
|
| 100 |
output = "原文: \n"
|
| 101 |
for index, sentence in enumerate(sentences):
|
| 102 |
output += f"{index}: {sentence}\n"
|
|
|
|
| 1 |
+
import math
|
| 2 |
+
|
| 3 |
import numpy
|
| 4 |
import torch
|
| 5 |
import gradio as gr
|
|
|
|
| 66 |
def __init__(self):
|
| 67 |
self.model = SentenceTransformer('paraphrase-multilingual-mpnet-base-v2')
|
| 68 |
self.ht = HarvestText()
|
| 69 |
+
|
| 70 |
+
def find_central(self, content: str, num=100):
|
| 71 |
sentences = self.ht.cut_sentences(content)
|
| 72 |
embeddings = self.model.encode(sentences, convert_to_tensor=True).cpu()
|
| 73 |
|
|
|
|
| 80 |
# We argsort so that the first element is the sentence with the highest score
|
| 81 |
most_central_sentence_indices = numpy.argsort(-centrality_scores)
|
| 82 |
|
| 83 |
+
# num = 100
|
| 84 |
res = []
|
| 85 |
for index in most_central_sentence_indices:
|
| 86 |
if num < 0:
|
|
|
|
| 99 |
|
| 100 |
|
| 101 |
def randeng_extract(content):
|
| 102 |
+
summary_length = math.ceil(len(content) / 10)
|
| 103 |
+
sentences = lex.find_central(content, num=summary_length)
|
| 104 |
output = "原文: \n"
|
| 105 |
for index, sentence in enumerate(sentences):
|
| 106 |
output += f"{index}: {sentence}\n"
|