Update app.py
Browse files
app.py
CHANGED
@@ -51,14 +51,14 @@ Here are some sample texts in Norwegian Bokmål and Norwegian Nynorsk that you c
|
|
51 |
|
52 |
masked_texts = [
|
53 |
"Hvordan kan man lære maskinen å oversette?",
|
54 |
-
"Ho vil ikkje gi bort
|
55 |
"Jeg tror vi har lært maskinene både nynorsk og bokmål",
|
56 |
"Aust er ikkje riktig retning.",
|
57 |
]
|
58 |
input_text = st.sidebar.selectbox("Select a Text", options=masked_texts)
|
59 |
|
60 |
st.sidebar.write("""
|
61 |
-
As you can see there are a lot of similarities between the languages. Since there also are some grammatical differences,
|
62 |
""")
|
63 |
|
64 |
text = st.text_area(" ",
|
|
|
51 |
|
52 |
masked_texts = [
|
53 |
"Hvordan kan man lære maskinen å oversette?",
|
54 |
+
"Ho vil ikkje gi bort dei personlege dataa sine.",
|
55 |
"Jeg tror vi har lært maskinene både nynorsk og bokmål",
|
56 |
"Aust er ikkje riktig retning.",
|
57 |
]
|
58 |
input_text = st.sidebar.selectbox("Select a Text", options=masked_texts)
|
59 |
|
60 |
st.sidebar.write("""
|
61 |
+
As you can see there are a lot of similarities between the languages. Since there also are some grammatical differences, simple dictionary replacements do not give a good result. A finetuned model on top of a pretrained t5-base from a balanced corpus, seem to solve the task with a SACREBLEU-score of 88.17.
|
62 |
""")
|
63 |
|
64 |
text = st.text_area(" ",
|