app.py

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)

@st.cache(allow_output_mutation=True, suppress_st_warning=True)
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 hash_func(inp):
    return True

@st.cache(allow_output_mutation=True, suppress_st_warning=True)
def loading_models(model='roberta-base'):
     return get_model(model), SentenceTransformer('all-MiniLM-L6-v2')

@st.cache(allow_output_mutation=True, 
          suppress_st_warning=True,
          hash_funcs={'tokenizers.Tokenizer': hash_func, 'tokenizers.AddedToken': hash_func})
def infer(text):
#    global nlp 
    return nlp(text+' '+nlp.tokenizer.mask_token)


@st.cache(allow_output_mutation=True, 
          suppress_st_warning=True,
          hash_funcs={'tokenizers.Tokenizer': hash_func, 'tokenizers.AddedToken': hash_func})
def sim(predicted_seq, sem_list):
    return semantic_model.encode(predicted_seq, convert_to_tensor=True), \
            semantic_model.encode(sem_list, convert_to_tensor=True)
    
@st.cache(allow_output_mutation=True, 
          suppress_st_warning=True,
          hash_funcs={'tokenizers.Tokenizer': hash_func, 'tokenizers.AddedToken': hash_func})
def main(text,semantic_text,history_keyword_text):
    global semantic_model, data_load_state
    data_load_state.text('Inference from model...')
    result = infer(text)
    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_history_embeddings = sim(predicted_seq, sem_list)
        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)
    return df
    
    
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 ...')

        nlp, semantic_model = loading_models(model)
        
        df=main(text,semantic_text,history_keyword_text)
        #show the results as a table
        st.table(df)
        data_load_state.text('')
    else:
        sys.argv = ['streamlit', 'run', sys.argv[0]]
        sys.exit(stcli.main())