Spaces:
Runtime error
Runtime error
| import streamlit as st | |
| import pandas as pd | |
| from streamlit import cli as stcli | |
| from transformers import pipeline | |
| from sentence_transformers import SentenceTransformer, util | |
| import sys | |
| HISTORY_WEIGHT = 100 # set history weight (if found any keyword from history, it will priorities based on its weight) | |
| def get_model(model): | |
| return pipeline("fill-mask", model=model, top_k=10)#set the maximum of tokens to be retrieved after each inference to model | |
| def loading_models(model='roberta-base'): | |
| return get_model(model), SentenceTransformer('all-MiniLM-L6-v2') | |
| def main(nlp, semantic_model, data_load_state): | |
| data_load_state.text('Inference from model...') | |
| result = nlp(text+' '+nlp.tokenizer.mask_token) | |
| sem_list=[semantic_text.strip()] | |
| data_load_state.text('Checking similarity...') | |
| if len(semantic_text): | |
| predicted_seq=[rec['sequence'] for rec in result] | |
| predicted_embeddings = semantic_model.encode(predicted_seq, convert_to_tensor=True) | |
| semantic_history_embeddings = semantic_model.encode(sem_list, convert_to_tensor=True) | |
| cosine_scores = util.cos_sim(predicted_embeddings, semantic_history_embeddings) | |
| data_load_state.text('similarity check completed...') | |
| for index, r in enumerate(result): | |
| if len(semantic_text): | |
| if len(r['token_str'])>2: #skip spcial chars such as "?" | |
| result[index]['score']+=float(sum(cosine_scores[index]))*HISTORY_WEIGHT | |
| if r['token_str'].lower().strip() in history_keyword_text.lower().strip() and len(r['token_str'].lower().strip())>1: | |
| #found from history, then increase the score of tokens | |
| result[index]['score']*=HISTORY_WEIGHT | |
| data_load_state.text('Score updated...') | |
| #sort the results | |
| df=pd.DataFrame(result).sort_values(by='score', ascending=False) | |
| # show the results as a table | |
| st.table(df) | |
| # print(df) | |
| data_load_state.text('') | |
| if __name__ == '__main__': | |
| if st._is_running_with_streamlit: | |
| st.markdown(""" | |
| # Auto-Complete | |
| This is an example of an auto-complete approach where the next token suggested based on users's history Keyword match & Semantic similarity of users's history (log). | |
| The next token is predicted per probability and a weight if it is appeared in keyword user's history or there is a similarity to semantic user's history | |
| """) | |
| history_keyword_text = st.text_input("Enter users's history <Keywords Match> (optional, i.e., 'Gates')", value="") | |
| semantic_text = st.text_input("Enter users's history <Semantic> (optional, i.e., 'Microsoft' or 'President')", value="Microsoft") | |
| text = st.text_input("Enter a text for auto completion...", value='Where is Bill') | |
| model = st.selectbox("Choose a model", ["roberta-base", "bert-base-uncased"]) | |
| data_load_state = st.text('1.Loading model ...') | |
| # semantic_model = SentenceTransformer('all-MiniLM-L6-v2') | |
| # nlp = get_model(model) | |
| nlp, semantic_model = loading_models(model) | |
| main(nlp, semantic_model, data_load_state) | |
| else: | |
| sys.argv = ['streamlit', 'run', sys.argv[0]] | |
| sys.exit(stcli.main()) |