Spaces:
Sleeping
Sleeping
Commit
·
fbb3b12
1
Parent(s):
a847a9a
Update app.py
Browse files
app.py
CHANGED
@@ -9,14 +9,14 @@ st.set_page_config(layout="wide")
|
|
9 |
with st.sidebar:
|
10 |
subset = st.selectbox('Flores eng_Latn-ukr_Cyrl subset', ('dev', 'devtest'))
|
11 |
|
12 |
-
tokenizer =
|
13 |
flores = load_dataset("facebook/flores", "eng_Latn-ukr_Cyrl")
|
14 |
dataset = flores[subset]
|
15 |
|
16 |
fig, (axl, axr) = plt.subplots(1, 2, figsize=(10,5))
|
17 |
-
axl.hist(
|
18 |
axl.set_title('eng mistral tokens')
|
19 |
-
axr.hist(
|
20 |
axr.set_title('ukr mistral tokens')
|
21 |
|
22 |
st.dataframe(pd.DataFrame(dataset[subset]))
|
|
|
9 |
with st.sidebar:
|
10 |
subset = st.selectbox('Flores eng_Latn-ukr_Cyrl subset', ('dev', 'devtest'))
|
11 |
|
12 |
+
tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-v0.1")
|
13 |
flores = load_dataset("facebook/flores", "eng_Latn-ukr_Cyrl")
|
14 |
dataset = flores[subset]
|
15 |
|
16 |
fig, (axl, axr) = plt.subplots(1, 2, figsize=(10,5))
|
17 |
+
axl.hist(dataset.map(lambda x: {'num_tokens':len(tok(x['sentence_eng_Latn'])['input_ids'])})['num_tokens'])
|
18 |
axl.set_title('eng mistral tokens')
|
19 |
+
axr.hist(dataset.map(lambda x: {'num_tokens':len(tok(x['sentence_ukr_Cyrl'])['input_ids'])})['num_tokens'])
|
20 |
axr.set_title('ukr mistral tokens')
|
21 |
|
22 |
st.dataframe(pd.DataFrame(dataset[subset]))
|