Spaces:
Running
on
Zero
Running
on
Zero
Peter
commited on
Commit
•
84b8f8b
1
Parent(s):
151824c
add onnx support
Browse filesSigned-off-by: Peter <74869040+pszemraj@users.noreply.github.com>
- app.py +39 -8
- requirements.txt +3 -1
app.py
CHANGED
@@ -1,13 +1,31 @@
|
|
1 |
import re
|
|
|
2 |
|
3 |
from cleantext import clean
|
4 |
import gradio as gr
|
5 |
from tqdm.auto import tqdm
|
6 |
from transformers import pipeline
|
7 |
|
|
|
|
|
|
|
|
|
8 |
# pipelines
|
9 |
-
checker = pipeline(
|
10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
|
12 |
def split_text(text: str) -> list:
|
13 |
# Split the text into sentences using regex
|
@@ -31,6 +49,7 @@ def split_text(text: str) -> list:
|
|
31 |
|
32 |
return sentence_batches
|
33 |
|
|
|
34 |
def correct_text(text: str, checker, corrector, separator: str = " ") -> str:
|
35 |
# Split the text into sentence batches
|
36 |
sentence_batches = split_text(text)
|
@@ -63,23 +82,35 @@ def correct_text(text: str, checker, corrector, separator: str = " ") -> str:
|
|
63 |
|
64 |
return corrected_text
|
65 |
|
|
|
66 |
def update(text: str):
|
67 |
text = clean(text[:4000], lower=False)
|
68 |
return correct_text(text, checker, corrector)
|
69 |
|
|
|
70 |
with gr.Blocks() as demo:
|
71 |
gr.Markdown("# <center>Robust Grammar Correction with FLAN-T5</center>")
|
72 |
-
gr.Markdown(
|
73 |
-
|
|
|
|
|
|
|
74 |
- `textattack/roberta-base-CoLA` for grammar quality detection
|
75 |
- `pszemraj/flan-t5-large-grammar-synthesis` for grammar correction
|
76 |
-
"""
|
|
|
77 |
with gr.Row():
|
78 |
-
inp = gr.Textbox(
|
|
|
|
|
|
|
|
|
79 |
out = gr.Textbox(label="output", interactive=False)
|
80 |
btn = gr.Button("Process")
|
81 |
btn.click(fn=update, inputs=inp, outputs=out)
|
82 |
gr.Markdown("---")
|
83 |
-
gr.Markdown(
|
|
|
|
|
84 |
gr.Markdown("- if experiencing long wait times, feel free to duplicate the space!")
|
85 |
-
demo.launch()
|
|
|
1 |
import re
|
2 |
+
import os
|
3 |
|
4 |
from cleantext import clean
|
5 |
import gradio as gr
|
6 |
from tqdm.auto import tqdm
|
7 |
from transformers import pipeline
|
8 |
|
9 |
+
|
10 |
+
checker_model_name = "textattack/roberta-base-CoLA"
|
11 |
+
corrector_model_name = "pszemraj/flan-t5-large-grammar-synthesis"
|
12 |
+
|
13 |
# pipelines
|
14 |
+
checker = pipeline(
|
15 |
+
"text-classification",
|
16 |
+
checker_model_name,
|
17 |
+
)
|
18 |
+
|
19 |
+
if os.environ.get("HF_DEMO_NO_USE_ONNX") is None:
|
20 |
+
# load onnx runtime unless HF_DEMO_NO_USE_ONNX is set
|
21 |
+
from optimum.pipelines import pipeline
|
22 |
+
|
23 |
+
corrector = pipeline(
|
24 |
+
"text2text-generation", model=corrector_model_name, accelerator="ort"
|
25 |
+
)
|
26 |
+
else:
|
27 |
+
corrector = pipeline("text2text-generation", corrector_model_name)
|
28 |
+
|
29 |
|
30 |
def split_text(text: str) -> list:
|
31 |
# Split the text into sentences using regex
|
|
|
49 |
|
50 |
return sentence_batches
|
51 |
|
52 |
+
|
53 |
def correct_text(text: str, checker, corrector, separator: str = " ") -> str:
|
54 |
# Split the text into sentence batches
|
55 |
sentence_batches = split_text(text)
|
|
|
82 |
|
83 |
return corrected_text
|
84 |
|
85 |
+
|
86 |
def update(text: str):
|
87 |
text = clean(text[:4000], lower=False)
|
88 |
return correct_text(text, checker, corrector)
|
89 |
|
90 |
+
|
91 |
with gr.Blocks() as demo:
|
92 |
gr.Markdown("# <center>Robust Grammar Correction with FLAN-T5</center>")
|
93 |
+
gr.Markdown(
|
94 |
+
"**Instructions:** Enter the text you want to correct in the textbox below (_text will be truncated to 4000 characters_). Click 'Process' to run."
|
95 |
+
)
|
96 |
+
gr.Markdown(
|
97 |
+
"""Models:
|
98 |
- `textattack/roberta-base-CoLA` for grammar quality detection
|
99 |
- `pszemraj/flan-t5-large-grammar-synthesis` for grammar correction
|
100 |
+
"""
|
101 |
+
)
|
102 |
with gr.Row():
|
103 |
+
inp = gr.Textbox(
|
104 |
+
label="input",
|
105 |
+
placeholder="PUT TEXT TO CHECK & CORRECT BROSKI",
|
106 |
+
value="I wen to the store yesturday to bye some food. I needd milk, bread, and a few otter things. The store was really crowed and I had a hard time finding everyting I needed. I finaly made it to the check out line and payed for my stuff.",
|
107 |
+
)
|
108 |
out = gr.Textbox(label="output", interactive=False)
|
109 |
btn = gr.Button("Process")
|
110 |
btn.click(fn=update, inputs=inp, outputs=out)
|
111 |
gr.Markdown("---")
|
112 |
+
gr.Markdown(
|
113 |
+
"- see the [model card](https://huggingface.co/pszemraj/flan-t5-large-grammar-synthesis) for more info"
|
114 |
+
)
|
115 |
gr.Markdown("- if experiencing long wait times, feel free to duplicate the space!")
|
116 |
+
demo.launch()
|
requirements.txt
CHANGED
@@ -2,4 +2,6 @@ transformers
|
|
2 |
gradio
|
3 |
tqdm
|
4 |
torch
|
5 |
-
clean-text
|
|
|
|
|
|
2 |
gradio
|
3 |
tqdm
|
4 |
torch
|
5 |
+
clean-text
|
6 |
+
accelerate
|
7 |
+
optimum[onnxruntime]
|