spuun commited on
Commit
5cce873
·
1 Parent(s): a19ef58

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +19 -19
app.py CHANGED
@@ -4,30 +4,30 @@ model_id = "knkarthick/MEETING_SUMMARY"
4
 
5
  generator = pipeline(task="text2text-generation", model=model_id)
6
 
7
- #def split_paragraph(paragraph, max_chunk_size=1024):
8
- # words = paragraph.split()
9
- # chunks = []
10
- # current_chunk = []
11
- # current_chunk_size = 0
12
 
13
- # for word in words:
14
- # word_len = len(word) + 1 # Add 1 for the space
15
- # if current_chunk_size + word_len <= max_chunk_size:
16
- # current_chunk.append(word)
17
- # current_chunk_size += word_len
18
- # else:
19
- # chunks.append(' '.join(current_chunk))
20
- # current_chunk = [word]
21
- # current_chunk_size = word_len
22
 
23
- # if current_chunk:
24
- # chunks.append(' '.join(current_chunk))
25
 
26
- # return chunks
27
 
28
  def launch(input):
29
- # if len(input) > 1024:
30
- # return " ".join([res["generated_text"] for res in generator(split_paragraph(input))])
31
  return generator(input)[0]["generated_text"]
32
 
33
  iface = gr.Interface(launch, inputs="text", outputs="text")
 
4
 
5
  generator = pipeline(task="text2text-generation", model=model_id)
6
 
7
+ def split_paragraph(paragraph, max_chunk_size=1024):
8
+ words = paragraph.split()
9
+ chunks = []
10
+ current_chunk = []
11
+ current_chunk_size = 0
12
 
13
+ for word in words:
14
+ word_len = len(word) + 1 # Add 1 for the space
15
+ if current_chunk_size + word_len <= max_chunk_size:
16
+ current_chunk.append(word)
17
+ current_chunk_size += word_len
18
+ else:
19
+ chunks.append(' '.join(current_chunk))
20
+ current_chunk = [word]
21
+ current_chunk_size = word_len
22
 
23
+ if current_chunk:
24
+ chunks.append(' '.join(current_chunk))
25
 
26
+ return chunks
27
 
28
  def launch(input):
29
+ if len(input) > 1024:
30
+ return " ".join([res["generated_text"] for res in generator(split_paragraph(input))])
31
  return generator(input)[0]["generated_text"]
32
 
33
  iface = gr.Interface(launch, inputs="text", outputs="text")