Spaces:
Build error
Build error
hellopahe commited on
Commit ·
f1ae6c0
1
Parent(s): 6a0cb69
add custom siblings
Browse files
app.py
CHANGED
|
@@ -1,7 +1,7 @@
|
|
| 1 |
import math, torch, gradio as gr
|
| 2 |
|
| 3 |
from lex_rank import LexRank
|
| 4 |
-
from
|
| 5 |
from lex_rank_L12 import LexRankL12
|
| 6 |
from sentence_transformers import SentenceTransformer, util
|
| 7 |
|
|
@@ -9,14 +9,14 @@ from sentence_transformers import SentenceTransformer, util
|
|
| 9 |
# ---===--- instances ---===---
|
| 10 |
embedder = SentenceTransformer('paraphrase-multilingual-mpnet-base-v2')
|
| 11 |
lex = LexRank()
|
| 12 |
-
lex_distiluse_v1 =
|
| 13 |
lex_l12 = LexRankL12()
|
| 14 |
|
| 15 |
|
| 16 |
# 摘要方法1
|
| 17 |
-
def extract_handler(content):
|
| 18 |
summary_length = math.ceil(len(content) / 10)
|
| 19 |
-
sentences = lex.find_central(content)
|
| 20 |
output = ""
|
| 21 |
for index, sentence in enumerate(sentences):
|
| 22 |
output += f"{index}: {sentence}\n"
|
|
@@ -24,9 +24,9 @@ def extract_handler(content):
|
|
| 24 |
|
| 25 |
|
| 26 |
# 摘要方法2
|
| 27 |
-
def extract_handler_distiluse_v1(content):
|
| 28 |
summary_length = math.ceil(len(content) / 10)
|
| 29 |
-
sentences = lex_distiluse_v1.find_central(content)
|
| 30 |
output = ""
|
| 31 |
for index, sentence in enumerate(sentences):
|
| 32 |
output += f"{index}: {sentence}\n"
|
|
@@ -34,9 +34,9 @@ def extract_handler_distiluse_v1(content):
|
|
| 34 |
|
| 35 |
|
| 36 |
# 摘要方法3
|
| 37 |
-
def extract_handler_l12(content):
|
| 38 |
summary_length = math.ceil(len(content) / 10)
|
| 39 |
-
sentences = lex_l12.find_central(content)
|
| 40 |
output = ""
|
| 41 |
for index, sentence in enumerate(sentences):
|
| 42 |
output += f"{index}: {sentence}\n"
|
|
@@ -69,16 +69,25 @@ with gr.Blocks() as app:
|
|
| 69 |
gr.Markdown("从下面的标签选择测试模块 [摘要生成,相似度检测]")
|
| 70 |
with gr.Tab("LexRank-mpnet"):
|
| 71 |
text_input_1 = gr.Textbox(label="请输入长文本:", lines=10, max_lines=1000)
|
| 72 |
-
|
| 73 |
-
|
| 74 |
-
|
|
|
|
|
|
|
|
|
|
| 75 |
text_input_2 = gr.Textbox(label="请输入长文本:", lines=10, max_lines=1000)
|
| 76 |
-
|
| 77 |
-
|
|
|
|
|
|
|
|
|
|
| 78 |
with gr.Tab("LexRank-MiniLM-L12-v2"):
|
| 79 |
text_input_3 = gr.Textbox(label="请输入长文本:", lines=10, max_lines=1000)
|
| 80 |
-
|
| 81 |
-
|
|
|
|
|
|
|
|
|
|
| 82 |
with gr.Tab("相似度检测"):
|
| 83 |
with gr.Row():
|
| 84 |
text_input_query = gr.Textbox(lines=10, label="查询文本")
|
|
@@ -86,9 +95,9 @@ with gr.Blocks() as app:
|
|
| 86 |
text_button_similarity = gr.Button("对比相似度")
|
| 87 |
text_output_similarity = gr.Textbox()
|
| 88 |
|
| 89 |
-
text_button_1.click(extract_handler, inputs=text_input_1, outputs=text_output_1)
|
| 90 |
-
text_button_2.click(extract_handler_distiluse_v1, inputs=text_input_2, outputs=text_output_2)
|
| 91 |
-
text_button_3.click(extract_handler_l12, inputs=text_input_3, outputs=text_output_3)
|
| 92 |
text_button_similarity.click(similarity_search, inputs=[text_input_query, text_input_doc], outputs=text_output_similarity)
|
| 93 |
|
| 94 |
app.launch(
|
|
|
|
| 1 |
import math, torch, gradio as gr
|
| 2 |
|
| 3 |
from lex_rank import LexRank
|
| 4 |
+
from lex_rank_text2vec_v1 import LexRankText2VecV1
|
| 5 |
from lex_rank_L12 import LexRankL12
|
| 6 |
from sentence_transformers import SentenceTransformer, util
|
| 7 |
|
|
|
|
| 9 |
# ---===--- instances ---===---
|
| 10 |
embedder = SentenceTransformer('paraphrase-multilingual-mpnet-base-v2')
|
| 11 |
lex = LexRank()
|
| 12 |
+
lex_distiluse_v1 = LexRankText2VecV1()
|
| 13 |
lex_l12 = LexRankL12()
|
| 14 |
|
| 15 |
|
| 16 |
# 摘要方法1
|
| 17 |
+
def extract_handler(content, siblings, num):
|
| 18 |
summary_length = math.ceil(len(content) / 10)
|
| 19 |
+
sentences = lex.find_central(content, siblings=siblings, num=num)
|
| 20 |
output = ""
|
| 21 |
for index, sentence in enumerate(sentences):
|
| 22 |
output += f"{index}: {sentence}\n"
|
|
|
|
| 24 |
|
| 25 |
|
| 26 |
# 摘要方法2
|
| 27 |
+
def extract_handler_distiluse_v1(content, siblings, num):
|
| 28 |
summary_length = math.ceil(len(content) / 10)
|
| 29 |
+
sentences = lex_distiluse_v1.find_central(content, siblings=siblings, num=num)
|
| 30 |
output = ""
|
| 31 |
for index, sentence in enumerate(sentences):
|
| 32 |
output += f"{index}: {sentence}\n"
|
|
|
|
| 34 |
|
| 35 |
|
| 36 |
# 摘要方法3
|
| 37 |
+
def extract_handler_l12(content, siblings, num):
|
| 38 |
summary_length = math.ceil(len(content) / 10)
|
| 39 |
+
sentences = lex_l12.find_central(content, siblings=siblings, num=num)
|
| 40 |
output = ""
|
| 41 |
for index, sentence in enumerate(sentences):
|
| 42 |
output += f"{index}: {sentence}\n"
|
|
|
|
| 69 |
gr.Markdown("从下面的标签选择测试模块 [摘要生成,相似度检测]")
|
| 70 |
with gr.Tab("LexRank-mpnet"):
|
| 71 |
text_input_1 = gr.Textbox(label="请输入长文本:", lines=10, max_lines=1000)
|
| 72 |
+
with gr.Row():
|
| 73 |
+
text_button_1 = gr.Button("生成摘要")
|
| 74 |
+
siblings_input_1 = gr.Textbox("请输入摘要的宽度半径, 默认为0, 即显示摘要本身.")
|
| 75 |
+
num_input_1 = gr.Textbox("摘要的条数, 默认10条")
|
| 76 |
+
text_output_1 = gr.Textbox(label="摘要文本", lines=10)
|
| 77 |
+
with gr.Tab("shibing624/text2vec-base-chinese-paraphrase"):
|
| 78 |
text_input_2 = gr.Textbox(label="请输入长文本:", lines=10, max_lines=1000)
|
| 79 |
+
with gr.Row():
|
| 80 |
+
text_button_2 = gr.Button("生成摘要")
|
| 81 |
+
siblings_input_2 = gr.Textbox("请输入摘要的宽度半径, 默认为0, 即显示摘要本身.")
|
| 82 |
+
num_input_2 = gr.Textbox("摘要的条数, 默认10条")
|
| 83 |
+
text_output_2 = gr.Textbox(label="摘要文本", lines=10)
|
| 84 |
with gr.Tab("LexRank-MiniLM-L12-v2"):
|
| 85 |
text_input_3 = gr.Textbox(label="请输入长文本:", lines=10, max_lines=1000)
|
| 86 |
+
with gr.Row():
|
| 87 |
+
text_button_3 = gr.Button("生成摘要")
|
| 88 |
+
siblings_input_3 = gr.Textbox("请输入摘要的宽度半径, 默认为0, 即显示摘要本身.")
|
| 89 |
+
num_input_3 = gr.Textbox("摘要的条数, 默认10条")
|
| 90 |
+
text_output_3 = gr.Textbox(label="摘要文本", lines=10)
|
| 91 |
with gr.Tab("相似度检测"):
|
| 92 |
with gr.Row():
|
| 93 |
text_input_query = gr.Textbox(lines=10, label="查询文本")
|
|
|
|
| 95 |
text_button_similarity = gr.Button("对比相似度")
|
| 96 |
text_output_similarity = gr.Textbox()
|
| 97 |
|
| 98 |
+
text_button_1.click(extract_handler, inputs=[text_input_1, siblings_input_1, num_input_1], outputs=text_output_1)
|
| 99 |
+
text_button_2.click(extract_handler_distiluse_v1, inputs=[text_input_2, siblings_input_2, num_input_2], outputs=text_output_2)
|
| 100 |
+
text_button_3.click(extract_handler_l12, inputs=[text_input_3, siblings_input_3, num_input_3], outputs=text_output_3)
|
| 101 |
text_button_similarity.click(similarity_search, inputs=[text_input_query, text_input_doc], outputs=text_output_similarity)
|
| 102 |
|
| 103 |
app.launch(
|