app.py

# import gradio as gr
import tensorflow as tf
import numpy as np
from keras.models import load_model
from tensorflow.keras.preprocessing.text import Tokenizer
import pickle
from tensorflow.keras.preprocessing.sequence import pad_sequences
import os
from pathlib import Path
import pandas as pd
import plotly.express as px
import keras
import unicodedata as ud

from underthesea import word_tokenize

from phoBERT import BERT_predict
import time

LSTM_model = tf.keras.models.load_model('lstm_model.tf')

GRU_model = tf.keras.models.load_model('gru_model.tf')

def LSTM_predict(x):
    t1 = time.time()
    pred_proba = LSTM_model.predict([x])[0]
    t2 = time.time()
    print(f'LSTM: {t2-t1}s')
    
    pred_proba = [round(i,2) for i in pred_proba]

    return pred_proba

def GRU_predict(x):
    t1 = time.time()
    pred_proba = GRU_model.predict([x])[0]
    t2 = time.time()
    print(f'GRU: {t2-t1}s')

    pred_proba = [round(i,2) for i in pred_proba]

    return pred_proba

def tokenize(x):
  x = ud.normalize('NFKC', x)
  x = word_tokenize(x, format="text")
  return x


def judge(x):

  result = []

  x = tokenize(x)

  lstm_pred = LSTM_predict(x)
  gru_pred = GRU_predict(x)
  
  result_lstm = np.round(lstm_pred, 2)
  result_gru = np.round(gru_pred, 2)

  for i in range(6):
    result.append((result_lstm[i]+result_gru[i])/2)
  
  return (result)


def judgePlus(x):

  result = []

  x = tokenize(x)

  lstm_pred = LSTM_predict(x)
  gru_pred = GRU_predict(x)
  try:
    bert_pred = BERT_predict(x)
  except:
    bert_pred = np.average([lstm_pred, gru_pred], axis=0)
  
  
  result_lstm = np.round(lstm_pred, 2)
  result_gru = np.round(gru_pred, 2)
  result_bert = np.round(bert_pred, 2)

  if((result_lstm[0]+result_gru[0])<(result_bert[0]*2)):
    for i in range(6):
      result.append((result_bert[i])/1)
  else:
    for i in range(6):
      result.append((result_lstm[i]+result_gru[i])/2)
    
  return (result)


def judgeBert(x):

  result = []

  x = tokenize(x)

  try:
    bert_pred = BERT_predict(x)
  except:
    bert_pred = np.zeros(6, dtype=float)
  
  result_bert = np.round(bert_pred, 2)
  
  for i in range(6):
    result.append((result_bert[i])/1)
  
  return (result)