Spaces:
Running
Running
Joshua Lochner
commited on
Commit
•
e68b946
1
Parent(s):
29f5ee8
Add option to streamlit app for model selection
Browse files
app.py
CHANGED
@@ -20,8 +20,8 @@ from evaluate import EvaluationArguments
|
|
20 |
from shared import device
|
21 |
|
22 |
st.set_page_config(
|
23 |
-
page_title=
|
24 |
-
page_icon=
|
25 |
# layout='wide',
|
26 |
# initial_sidebar_state="expanded",
|
27 |
menu_items={
|
@@ -30,8 +30,33 @@ st.set_page_config(
|
|
30 |
# 'About': "# This is a header. This is an *extremely* cool app!"
|
31 |
}
|
32 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
|
34 |
-
|
|
|
|
|
|
|
|
|
35 |
|
36 |
CLASSIFIER_PATH = 'Xenova/sponsorblock-classifier'
|
37 |
|
@@ -46,9 +71,9 @@ predictions_cache = persistdata()
|
|
46 |
|
47 |
|
48 |
@st.cache(allow_output_mutation=True)
|
49 |
-
def load_predict():
|
50 |
# Use default segmentation and classification arguments
|
51 |
-
evaluation_args = EvaluationArguments(model_path=
|
52 |
segmentation_args = SegmentationArguments()
|
53 |
classifier_args = ClassifierArguments()
|
54 |
|
@@ -81,24 +106,17 @@ def load_predict():
|
|
81 |
return predict_function
|
82 |
|
83 |
|
84 |
-
CATGEGORY_OPTIONS = {
|
85 |
-
'SPONSOR': 'Sponsor',
|
86 |
-
'SELFPROMO': 'Self/unpaid promo',
|
87 |
-
'INTERACTION': 'Interaction reminder',
|
88 |
-
}
|
89 |
-
|
90 |
-
|
91 |
-
# Load prediction function
|
92 |
-
predict = load_predict()
|
93 |
-
|
94 |
-
|
95 |
def main():
|
96 |
|
97 |
# Display heading and subheading
|
98 |
st.write('# SponsorBlock ML')
|
99 |
st.write('##### Automatically detect in-video YouTube sponsorships, self/unpaid promotions, and interaction reminders.')
|
100 |
|
101 |
-
|
|
|
|
|
|
|
|
|
102 |
video_id = st.text_input('Video ID:') # , placeholder='e.g., axtQvkSpoto'
|
103 |
|
104 |
categories = st.multiselect('Categories:',
|
|
|
20 |
from shared import device
|
21 |
|
22 |
st.set_page_config(
|
23 |
+
page_title='SponsorBlock ML',
|
24 |
+
page_icon='🤖',
|
25 |
# layout='wide',
|
26 |
# initial_sidebar_state="expanded",
|
27 |
menu_items={
|
|
|
30 |
# 'About': "# This is a header. This is an *extremely* cool app!"
|
31 |
}
|
32 |
)
|
33 |
+
# https://github.com/google-research/text-to-text-transfer-transformer#released-model-checkpoints
|
34 |
+
# https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#experimental-t5-pre-trained-model-checkpoints
|
35 |
+
|
36 |
+
# https://huggingface.co/docs/transformers/model_doc/t5
|
37 |
+
# https://huggingface.co/docs/transformers/model_doc/t5v1.1
|
38 |
+
MODELS = {
|
39 |
+
'Small (77M)': {
|
40 |
+
'pretrained': 'google/t5-v1_1-small',
|
41 |
+
'repo_id': 'Xenova/sponsorblock-small',
|
42 |
+
},
|
43 |
+
'Base v1 (220M)': {
|
44 |
+
'pretrained': 't5-base',
|
45 |
+
'repo_id': 'EColi/sponsorblock-base-v1',
|
46 |
+
},
|
47 |
+
|
48 |
+
'Base v1.1 (250M)': {
|
49 |
+
'pretrained': 'google/t5-v1_1-base',
|
50 |
+
'repo_id': 'Xenova/sponsorblock-base',
|
51 |
+
}
|
52 |
+
|
53 |
+
}
|
54 |
|
55 |
+
CATGEGORY_OPTIONS = {
|
56 |
+
'SPONSOR': 'Sponsor',
|
57 |
+
'SELFPROMO': 'Self/unpaid promo',
|
58 |
+
'INTERACTION': 'Interaction reminder',
|
59 |
+
}
|
60 |
|
61 |
CLASSIFIER_PATH = 'Xenova/sponsorblock-classifier'
|
62 |
|
|
|
71 |
|
72 |
|
73 |
@st.cache(allow_output_mutation=True)
|
74 |
+
def load_predict(model_path):
|
75 |
# Use default segmentation and classification arguments
|
76 |
+
evaluation_args = EvaluationArguments(model_path=model_path)
|
77 |
segmentation_args = SegmentationArguments()
|
78 |
classifier_args = ClassifierArguments()
|
79 |
|
|
|
106 |
return predict_function
|
107 |
|
108 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
109 |
def main():
|
110 |
|
111 |
# Display heading and subheading
|
112 |
st.write('# SponsorBlock ML')
|
113 |
st.write('##### Automatically detect in-video YouTube sponsorships, self/unpaid promotions, and interaction reminders.')
|
114 |
|
115 |
+
model_id = st.selectbox('Select model', MODELS.keys(), index=0)
|
116 |
+
|
117 |
+
# Load prediction function
|
118 |
+
predict = load_predict(MODELS[model_id]['repo_id'])
|
119 |
+
|
120 |
video_id = st.text_input('Video ID:') # , placeholder='e.g., axtQvkSpoto'
|
121 |
|
122 |
categories = st.multiselect('Categories:',
|