app.py

import streamlit as st
from utils import get_res


st.sidebar.title('Tokenizers demo')

#x = st.slider('Select a value')
#st.write(x, 'squared is', x * x)

model_option = ['deepseek-ai/deepseek-coder-1.3b-instruct',
             'MediaTek-Research/Breeze-7B-Instruct-64k-v0_1',
             'microsoft/phi-2', 'bigcode/starcoder', 'enter by myself']

input_option = ['123.5', 'hello world!!!', '大雨＋寒流來襲！全台極凍72小時「探5度以下', 'enter by myself']


st.sidebar.subheader('Choose the tokenizer', divider='grey')
st.sidebar.write('You can choose `enter by myself` to paste the model you want.')
model_name_A = st.sidebar.selectbox(
    'Model Name A',
    model_option)


if model_name_A == 'enter by myself':
    model_name_A = st.sidebar.text_input('Please enter Model Name A', 'deepseek-ai/deepseek-coder-1.3b-instruct')

model_name_B = st.sidebar.selectbox(
    'Model Name B',
    model_option)

if model_name_B == 'enter by myself':
    model_name_B = st.sidebar.text_input('Please enter Model Name B', 'deepseek-ai/deepseek-coder-1.3b-instruct')



#with st.sidebar.expander("Models that you might want"):
#    for m in model_option:
#        st.write(m)


#'Your choice:', model_name

st.sidebar.subheader('Choose the input sentence', divider='grey')
st.sidebar.write('You can choose `enter by myself` to enter the text you want.')
input_data = st.sidebar.selectbox(
    'Input Sentence',
    input_option)

if input_data == 'enter by myself':
    input_data = st.sidebar.text_input('Write the Input Sentence', 'Hello sunshine!!!')

#with st.sidebar.expander("Input that you might want to test"):
#    for m in input_option:
#        st.write(m)


col1, col2 = st.columns(2)

with col1:
    st.subheader(model_name_A, divider='grey')
    res, token_num = get_res(model_name=model_name_A, input_sentence=input_data, single_print=False)

    st.subheader('Tokenized result')
    st.markdown(res, unsafe_allow_html=True)

    st.subheader('Number of tokens')
    st.markdown(f'<span style="font-size:1.875em">{str(token_num)}</span>', 
                unsafe_allow_html=True)

with col2:
    st.subheader(model_name_B, divider='grey')
    res, token_num = get_res(model_name=model_name_B, input_sentence=input_data, single_print=False)

    st.subheader('Tokenized result')
    st.markdown(res, unsafe_allow_html=True)

    st.subheader('Number of tokens')
    st.markdown(f'<span style="font-size:1.875em">{str(token_num)}</span>', 
                unsafe_allow_html=True)