jafrilalam
commited on
Update app.py
Browse files
app.py
CHANGED
@@ -1,3 +1,29 @@
|
|
1 |
import gradio as gr
|
|
|
2 |
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import gradio as gr
|
2 |
+
from transformers import pipeline, AutoModelForSeq2SeqLM, MBart50Tokenizer, AutoTokenizer
|
3 |
|
4 |
+
tokenizer = AutoTokenizer.from_pretrained('jafrilalam/bangla_sentence_correction_01', src_lang="bn_IN", tgt_lang="bn_IN", use_fast=True)
|
5 |
+
model = AutoModelForSeq2SeqLM.from_pretrained("jafrilalam/bangla_sentence_correction_01", use_safetensors=True)
|
6 |
+
|
7 |
+
def correct_text(given_sentence):
|
8 |
+
inputs = tokenizer.encode(
|
9 |
+
given_sentence,
|
10 |
+
truncation=True,
|
11 |
+
return_tensors="pt",
|
12 |
+
max_length=len(given_sentence),
|
13 |
+
)
|
14 |
+
|
15 |
+
output_ids = model.generate(
|
16 |
+
inputs,
|
17 |
+
max_new_tokens=len(given_sentence),
|
18 |
+
early_stopping=True,
|
19 |
+
)
|
20 |
+
|
21 |
+
return tokenizer.decode(output_ids[0], skip_special_tokens=True)
|
22 |
+
|
23 |
+
iface = gr.Interface(
|
24 |
+
fn=correct_text,
|
25 |
+
inputs=gr.Textbox(lines=4, label="Incorrect Bangla Sentence"),
|
26 |
+
outputs=gr.Textbox(label="Corrected Bangla Sentence")
|
27 |
+
)
|
28 |
+
|
29 |
+
iface.launch(share = True)
|