Spaces:
Sleeping
Sleeping
fvancesco
commited on
Commit
·
e08b566
1
Parent(s):
f14078e
add description
Browse files
app.py
CHANGED
@@ -105,6 +105,24 @@ def add_mask(text):
|
|
105 |
|
106 |
with gr.Blocks() as demo:
|
107 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
108 |
textbox = gr.Textbox(value="Happy <mask>!", max_lines=1)
|
109 |
|
110 |
with gr.Row():
|
|
|
105 |
|
106 |
with gr.Blocks() as demo:
|
107 |
|
108 |
+
text_description="""
|
109 |
+
# TimeLMs Demo
|
110 |
+
|
111 |
+
This is a demo for **timeLMs**:
|
112 |
+
- [Github](https://github.com/cardiffnlp/timelms)
|
113 |
+
- [Paper](https://aclanthology.org/2022.acl-demo.25.pdf)
|
114 |
+
|
115 |
+
Input any text with a *\<mask\>* token as in the example, and (the demo does not
|
116 |
+
use GPUs, and it takes about 1 min). In the graph, we show the probability of
|
117 |
+
some token candidates for mask over different months.
|
118 |
+
|
119 |
+
In this demo we run use a roberta-base model trained on tweets, where the first two
|
120 |
+
tokens are the year and the month ("21 1" for January 2021). It was trained
|
121 |
+
for tweets between January 2018 to December 2021).
|
122 |
+
"""
|
123 |
+
|
124 |
+
description = gr.Markdown(text_description)
|
125 |
+
|
126 |
textbox = gr.Textbox(value="Happy <mask>!", max_lines=1)
|
127 |
|
128 |
with gr.Row():
|