Spaces:
Running
on
Zero
Running
on
Zero
Commit
·
82b0ab8
1
Parent(s):
8a588ad
- app.py +9 -5
- title_gen.py +4 -2
- typo_check.py +15 -8
app.py
CHANGED
@@ -1,7 +1,7 @@
|
|
1 |
import gradio as gr
|
2 |
|
3 |
-
from typo_check import css, process_input
|
4 |
-
from title_gen import generate_title,
|
5 |
|
6 |
|
7 |
# Create Gradio interface using the latest syntax
|
@@ -11,6 +11,9 @@ with gr.Blocks(theme=gr.themes.Default(), css=css) as demo:
|
|
11 |
gr.Markdown("# <center>Dhivehi Typo Correction</center>")
|
12 |
gr.Markdown("This app uses a fine-tuned T5 model to correct typos in Dhivehi text. Enter text with typos and the model will attempt to fix them.")
|
13 |
|
|
|
|
|
|
|
14 |
with gr.Row():
|
15 |
input_text = gr.Textbox(
|
16 |
lines=1,
|
@@ -19,7 +22,8 @@ with gr.Blocks(theme=gr.themes.Default(), css=css) as demo:
|
|
19 |
rtl=True,
|
20 |
elem_classes="textbox1"
|
21 |
)
|
22 |
-
|
|
|
23 |
with gr.Row():
|
24 |
corrected_text = gr.Textbox(
|
25 |
lines=1,
|
@@ -37,7 +41,7 @@ with gr.Blocks(theme=gr.themes.Default(), css=css) as demo:
|
|
37 |
submit_btn = gr.Button("ރަނގަޅު ކޮށްލުމަށް",elem_classes="textbox1") # "Correct" in Dhivehi
|
38 |
submit_btn.click(
|
39 |
fn=process_input,
|
40 |
-
inputs=input_text,
|
41 |
outputs=[corrected_text, highlighted_diff]
|
42 |
)
|
43 |
|
@@ -78,7 +82,7 @@ with gr.Blocks(theme=gr.themes.Default(), css=css) as demo:
|
|
78 |
with gr.Row():
|
79 |
article_content = gr.Textbox(lines=10, label="Article Content", rtl=True, elem_classes="textbox1")
|
80 |
with gr.Row():
|
81 |
-
model_choice = gr.Dropdown(choices=list(
|
82 |
with gr.Row():
|
83 |
seed = gr.Slider(0, 10000, value=42, step=1, label="Random Seed")
|
84 |
use_sampling = gr.Checkbox(label="Use Sampling (Creative/Random)", value=False)
|
|
|
1 |
import gradio as gr
|
2 |
|
3 |
+
from typo_check import css, process_input,MODEL_OPTIONS_TYPO
|
4 |
+
from title_gen import generate_title, MODEL_OPTIONS_TITLE
|
5 |
|
6 |
|
7 |
# Create Gradio interface using the latest syntax
|
|
|
11 |
gr.Markdown("# <center>Dhivehi Typo Correction</center>")
|
12 |
gr.Markdown("This app uses a fine-tuned T5 model to correct typos in Dhivehi text. Enter text with typos and the model will attempt to fix them.")
|
13 |
|
14 |
+
with gr.Row():
|
15 |
+
model_choice = gr.Dropdown(choices=list(MODEL_OPTIONS_TYPO.keys()), value="A3 Model", label="Model")
|
16 |
+
|
17 |
with gr.Row():
|
18 |
input_text = gr.Textbox(
|
19 |
lines=1,
|
|
|
22 |
rtl=True,
|
23 |
elem_classes="textbox1"
|
24 |
)
|
25 |
+
|
26 |
+
|
27 |
with gr.Row():
|
28 |
corrected_text = gr.Textbox(
|
29 |
lines=1,
|
|
|
41 |
submit_btn = gr.Button("ރަނގަޅު ކޮށްލުމަށް",elem_classes="textbox1") # "Correct" in Dhivehi
|
42 |
submit_btn.click(
|
43 |
fn=process_input,
|
44 |
+
inputs=[input_text,model_choice],
|
45 |
outputs=[corrected_text, highlighted_diff]
|
46 |
)
|
47 |
|
|
|
82 |
with gr.Row():
|
83 |
article_content = gr.Textbox(lines=10, label="Article Content", rtl=True, elem_classes="textbox1")
|
84 |
with gr.Row():
|
85 |
+
model_choice = gr.Dropdown(choices=list(MODEL_OPTIONS_TITLE.keys()), value="V6 Model", label="Model")
|
86 |
with gr.Row():
|
87 |
seed = gr.Slider(0, 10000, value=42, step=1, label="Random Seed")
|
88 |
use_sampling = gr.Checkbox(label="Use Sampling (Creative/Random)", value=False)
|
title_gen.py
CHANGED
@@ -2,9 +2,10 @@ import random
|
|
2 |
import numpy as np
|
3 |
import torch
|
4 |
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
|
|
5 |
|
6 |
# Available models
|
7 |
-
|
8 |
"V6 Model": "alakxender/t5-divehi-title-generation-v6",
|
9 |
"XS Model": "alakxender/t5-dhivehi-title-generation-xs"
|
10 |
}
|
@@ -28,6 +29,7 @@ prefix = "2title: "
|
|
28 |
max_input_length = 512
|
29 |
max_target_length = 32
|
30 |
|
|
|
31 |
def generate_title(content, seed, use_sampling, model_choice):
|
32 |
random.seed(seed)
|
33 |
np.random.seed(seed)
|
@@ -35,7 +37,7 @@ def generate_title(content, seed, use_sampling, model_choice):
|
|
35 |
if torch.cuda.is_available():
|
36 |
torch.cuda.manual_seed_all(seed)
|
37 |
|
38 |
-
model_dir =
|
39 |
tokenizer, model = get_model_and_tokenizer(model_dir)
|
40 |
|
41 |
input_text = prefix + content.strip()
|
|
|
2 |
import numpy as np
|
3 |
import torch
|
4 |
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
5 |
+
import spaces
|
6 |
|
7 |
# Available models
|
8 |
+
MODEL_OPTIONS_TITLE = {
|
9 |
"V6 Model": "alakxender/t5-divehi-title-generation-v6",
|
10 |
"XS Model": "alakxender/t5-dhivehi-title-generation-xs"
|
11 |
}
|
|
|
29 |
max_input_length = 512
|
30 |
max_target_length = 32
|
31 |
|
32 |
+
@spaces.GPU()
|
33 |
def generate_title(content, seed, use_sampling, model_choice):
|
34 |
random.seed(seed)
|
35 |
np.random.seed(seed)
|
|
|
37 |
if torch.cuda.is_available():
|
38 |
torch.cuda.manual_seed_all(seed)
|
39 |
|
40 |
+
model_dir = MODEL_OPTIONS_TITLE[model_choice]
|
41 |
tokenizer, model = get_model_and_tokenizer(model_dir)
|
42 |
|
43 |
input_text = prefix + content.strip()
|
typo_check.py
CHANGED
@@ -5,19 +5,25 @@ import difflib
|
|
5 |
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
6 |
import torch
|
7 |
import gradio as gr
|
|
|
8 |
|
9 |
-
#
|
10 |
-
|
|
|
|
|
|
|
11 |
|
12 |
# Function to load model and tokenizer
|
13 |
-
def load_model():
|
14 |
print("Loading model and tokenizer...")
|
15 |
try:
|
16 |
-
|
|
|
|
|
17 |
if tokenizer.pad_token is None:
|
18 |
tokenizer.pad_token = tokenizer.eos_token
|
19 |
|
20 |
-
model = AutoModelForSeq2SeqLM.from_pretrained(
|
21 |
|
22 |
# Move model to GPU if available
|
23 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
@@ -67,7 +73,7 @@ def correct_typo(text, model, tokenizer, device):
|
|
67 |
return f"Error: {str(e)}"
|
68 |
|
69 |
# Initialize model and tokenizer
|
70 |
-
model, tokenizer, device = load_model()
|
71 |
|
72 |
if model is None:
|
73 |
print("Failed to load model. Please check your model and tokenizer paths.")
|
@@ -103,9 +109,10 @@ def highlight_differences(original, corrected):
|
|
103 |
return f'<div class="dhivehi-diff">{" ".join(html_parts)}</div>'
|
104 |
|
105 |
# Function to process the input for Gradio
|
106 |
-
|
|
|
107 |
if model is None:
|
108 |
-
load_model()
|
109 |
|
110 |
corrected = correct_typo(text, model, tokenizer, device)
|
111 |
highlighted = highlight_differences(text, corrected)
|
|
|
5 |
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
6 |
import torch
|
7 |
import gradio as gr
|
8 |
+
import spaces
|
9 |
|
10 |
+
# Available models
|
11 |
+
MODEL_OPTIONS_TYPO = {
|
12 |
+
"A3 Model": "alakxender/t5-dhivehi-typo-corrector-asr",
|
13 |
+
"XS Model": "alakxender/dhivehi-quick-spell-check-t5"
|
14 |
+
}
|
15 |
|
16 |
# Function to load model and tokenizer
|
17 |
+
def load_model(model_choice):
|
18 |
print("Loading model and tokenizer...")
|
19 |
try:
|
20 |
+
selected_model = MODEL_OPTIONS_TYPO[model_choice]
|
21 |
+
|
22 |
+
tokenizer = AutoTokenizer.from_pretrained(selected_model)
|
23 |
if tokenizer.pad_token is None:
|
24 |
tokenizer.pad_token = tokenizer.eos_token
|
25 |
|
26 |
+
model = AutoModelForSeq2SeqLM.from_pretrained(selected_model)
|
27 |
|
28 |
# Move model to GPU if available
|
29 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
|
|
73 |
return f"Error: {str(e)}"
|
74 |
|
75 |
# Initialize model and tokenizer
|
76 |
+
model, tokenizer, device = load_model("A3 Model")
|
77 |
|
78 |
if model is None:
|
79 |
print("Failed to load model. Please check your model and tokenizer paths.")
|
|
|
109 |
return f'<div class="dhivehi-diff">{" ".join(html_parts)}</div>'
|
110 |
|
111 |
# Function to process the input for Gradio
|
112 |
+
@spaces.GPU()
|
113 |
+
def process_input(text,model_choice):
|
114 |
if model is None:
|
115 |
+
load_model(model_choice)
|
116 |
|
117 |
corrected = correct_typo(text, model, tokenizer, device)
|
118 |
highlighted = highlight_differences(text, corrected)
|