app.py

import pickle
import gradio as gr
import tensorflow.compat.v1 as tf
tf.disable_v2_behavior()

def transliterate_eng_hin(input_word):
    #loading the saved parameters
    def load_params():
        with open('./params.p', mode='rb') as in_file:
            return pickle.load(in_file)
    
    def load_preprocess():
        with open('./preprocess.p', mode='rb') as in_file:
            return pickle.load(in_file)
    
    #getting the source and target vocabuaries
    _, (source_vocab_to_int, target_vocab_to_int), (source_int_to_vocab, target_int_to_vocab) = load_preprocess()
    
    load_path = load_params()
    print("Loaded path:", load_path)
    print(type(load_path))
    
    batch_size = 30
    
    #converting the words to vectors of integers
    def word_to_seq(word, vocab_to_int):
        results = []
        for word in list(word):
            if word in vocab_to_int:
                results.append(vocab_to_int[word])
            else:
                results.append(vocab_to_int['<UNK>'])
                
        return results
    
    #taking user input for prediction
    # print("\n Enter word to be transliterated:")
    transliterate_word = input_word.lower()
    
    transliterate_word = word_to_seq(transliterate_word, source_vocab_to_int)
    
    #initialising the graph
    loaded_graph = tf.Graph()
    
    #initialising the session
    tf.compat.v1.Session()
    # with tf.Session(graph=loaded_graph) as sess:
    with tf.compat.v1.Session(graph=loaded_graph) as sess:
            
        # Load saved model
        loader = tf.train.import_meta_graph("./dev.meta")
        
    #     tf.train.Saver.restore(sess,load_path)
        loader.restore(sess, "./dev")
    
    #providing placeholder names from the loaded graph
        input_data = loaded_graph.get_tensor_by_name('input:0')
        logits = loaded_graph.get_tensor_by_name('predictions:0')
        target_sequence_length = loaded_graph.get_tensor_by_name('target_sequence_length:0')
        keep_prob = loaded_graph.get_tensor_by_name('keep_prob:0')
    
    #transliterating the given word
        transliterate_logits = sess.run(logits, {input_data: [transliterate_word]*batch_size,
                                             target_sequence_length: [len(transliterate_word)]*batch_size,
                                             keep_prob: 1.0})[0]
    
    print('Input')
    print('  Word Ids:      {}'.format([i for i in transliterate_word]))
    print('  English Word: {}'.format([source_int_to_vocab[i] for i in transliterate_word]))
    
    print('\nPrediction')
    print('  Word Id:      {}'.format([i for i in transliterate_logits]))
    
    #showing the output
    output = ""
    for i in transliterate_logits:
            if target_int_to_vocab[i]!= '<EOS>':
                    output = output + target_int_to_vocab[i]
    print('  Hindi Word:      {}'.format(output))

    return output

demo = gr.Interface(fn=transliterate_eng_hin, inputs="text", outputs="text")

if __name__ == "__main__":
    demo.launch()