from datasets import load_dataset
import pandas as pd
import streamlit as st
from transformers import AutoTokenizer
import matplotlib.pyplot as plt

st.set_page_config(layout="wide")

with st.sidebar:
    subset = st.selectbox('Flores eng_Latn-ukr_Cyrl subset', ('dev', 'devtest'))

tokenizer = transformers.AutoTokenizer.from_pretrained("mistralai/Mistral-7B-v0.1")
flores =  load_dataset("facebook/flores", "eng_Latn-ukr_Cyrl")
dataset = flores[subset]

fig, (axl, axr) = plt.subplots(1, 2, figsize=(10,5))
axl.hist(dataaset.map(lambda x: {'num_tokens':len(tok(x['sentence_eng_Latn'])['input_ids'])})['num_tokens'])
axl.set_title('eng mistral tokens')
axr.hist(dataaset.map(lambda x: {'num_tokens':len(tok(x['sentence_ukr_Cyrl'])['input_ids'])})['num_tokens'])
axr.set_title('ukr mistral tokens')

st.dataframe(pd.DataFrame(dataset[subset]))