Update app.py
Browse files
app.py
CHANGED
@@ -1,27 +1,49 @@
|
|
1 |
import gradio as gr
|
2 |
import sys
|
3 |
-
import
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4 |
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
|
12 |
def evaluate_translation(src_text, mt_text):
|
13 |
if not hasattr(evaluate_translation, "model"):
|
14 |
-
|
15 |
-
|
16 |
-
evaluate_translation.model = load_from_checkpoint(model_path)
|
17 |
-
except Exception as e:
|
18 |
-
return f"Error loading model: {str(e)}"
|
19 |
|
20 |
translations = [{"src": src_text, "mt": mt_text}]
|
21 |
results = evaluate_translation.model.predict(
|
22 |
translations,
|
23 |
batch_size=1,
|
24 |
-
gpus=0
|
25 |
)
|
26 |
return float(results['scores'][0])
|
27 |
|
|
|
1 |
import gradio as gr
|
2 |
import sys
|
3 |
+
import subprocess
|
4 |
+
import pkg_resources
|
5 |
+
|
6 |
+
def install_package(package):
|
7 |
+
try:
|
8 |
+
subprocess.check_call([sys.executable, "-m", "pip", "install", package])
|
9 |
+
return True
|
10 |
+
except:
|
11 |
+
return False
|
12 |
+
|
13 |
+
# Check and install missing dependencies
|
14 |
+
required_packages = ["torch", "comet"]
|
15 |
+
missing_packages = []
|
16 |
|
17 |
+
for package in required_packages:
|
18 |
+
try:
|
19 |
+
pkg_resources.get_distribution(package)
|
20 |
+
except pkg_resources.DistributionNotFound:
|
21 |
+
missing_packages.append(package)
|
22 |
+
|
23 |
+
if missing_packages:
|
24 |
+
print(f"Missing packages: {', '.join(missing_packages)}")
|
25 |
+
for package in missing_packages:
|
26 |
+
if install_package(package):
|
27 |
+
print(f"Successfully installed {package}")
|
28 |
+
else:
|
29 |
+
print(f"Failed to install {package}")
|
30 |
+
print(f"Please install manually: pip install {' '.join(required_packages)}")
|
31 |
+
sys.exit(1)
|
32 |
+
|
33 |
+
# Now import torch and comet after ensuring they're installed
|
34 |
+
import torch
|
35 |
+
from comet import download_model, load_from_checkpoint
|
36 |
|
37 |
def evaluate_translation(src_text, mt_text):
|
38 |
if not hasattr(evaluate_translation, "model"):
|
39 |
+
model_path = download_model("wasanx/ComeTH")
|
40 |
+
evaluate_translation.model = load_from_checkpoint(model_path)
|
|
|
|
|
|
|
41 |
|
42 |
translations = [{"src": src_text, "mt": mt_text}]
|
43 |
results = evaluate_translation.model.predict(
|
44 |
translations,
|
45 |
batch_size=1,
|
46 |
+
gpus=0
|
47 |
)
|
48 |
return float(results['scores'][0])
|
49 |
|