Spaces:
Runtime error
Runtime error
| import re | |
| import streamlit as st | |
| from wordcloud import WordCloud | |
| from googletrans import Translator | |
| from transformers import AutoTokenizer, AutoModelWithLMHead, pipeline | |
| from enums import MODEL_NAME, MESSAGES, DESCRIPTION | |
| def iso2lang(iso): | |
| return MESSAGES["iso"][iso] | |
| def simple_clean(text): | |
| return re.sub('[!@#$.,\n-?]', ' ', text.lower()) | |
| def load_tokenizer(): | |
| return AutoTokenizer.from_pretrained(MODEL_NAME) | |
| def load_model(): | |
| return AutoModelWithLMHead.from_pretrained(MODEL_NAME) | |
| def load_pipe(): | |
| model = load_model() | |
| tokenizer = load_tokenizer() | |
| return pipeline("text-generation", model=model, tokenizer=tokenizer) | |
| # ---------------------------------------------------------------------- # | |
| st.write(DESCRIPTION) | |
| lang = st.radio('Хэл / Language', ('mn', 'en'), format_func=iso2lang) | |
| translator = Translator() | |
| with st.spinner(MESSAGES["loading_text"][lang]): | |
| pipe = load_pipe() | |
| st.success(MESSAGES["success_model_load"][lang]) | |
| text = st.text_input( | |
| MESSAGES["input_description"][lang], MESSAGES["input_default"][lang]) | |
| with st.spinner(MESSAGES["loading_text"][lang]): | |
| if lang == "mn": | |
| result = pipe(text)[0]['generated_text'] | |
| st.write(result) | |
| elif lang == "en": | |
| text = translator.translate(text, src='en', dest='mn').text | |
| result = pipe(text)[0]['generated_text'] | |
| result_en = translator.translate(result, src='mn', dest='en').text | |
| st.write(f"*Translated:* {result_en}") | |
| st.write(f"> *Original:* {result}") | |
| st.warning('Translation is done by [`googletrans`](https://github.com/ssut/py-googletrans). Please check out the usage. https://github.com/ssut/py-googletrans#note-on-library-usage') | |
| st.write("### WordCloud:") | |
| with st.spinner(MESSAGES["loading_text"][lang]): | |
| wordcloud = WordCloud(width = 300, height = 300, | |
| background_color ='white', | |
| min_font_size = 5).generate(result) | |
| st.image(wordcloud.to_array()) | |