Spaces:
Runtime error
Runtime error
| #import pickle | |
| import gradio as gr | |
| import tensorflow as tf | |
| import numpy as np | |
| import json | |
| #import nltk | |
| #from sklearn.model_selection import train_test_split | |
| #from sklearn.preprocessing import LabelEncoder | |
| from tensorflow import keras | |
| from keras.preprocessing.text import Tokenizer | |
| from keras.models import load_model | |
| #from keras.models import Sequential | |
| #from keras.layers import Embedding, LSTM, Dense | |
| from tensorflow.keras.preprocessing.sequence import pad_sequences | |
| #from nltk.tokenize.simple import CharTokenizer | |
| tokenizer = Tokenizer(char_level=True) | |
| tokenizer_json = "tokenizer.json" # Replace with the path to your tokenizer configuration | |
| # Load the tokenizer configuration from the JSON file | |
| with open(tokenizer_json, "r") as json_file: | |
| tokenizer_config = json.load(json_file) | |
| # Manually set word_index and index_word from the 'config' dictionary | |
| tokenizer.word_index = tokenizer_config['config']['word_index'] | |
| tokenizer.index_word = tokenizer_config['config']['index_word'] | |
| #with open("word_guessing.pkl", "rb") as f: | |
| # model_standardform = pickle.load(f) | |
| model_standardform = keras.models.load_model("FINALword_guessing.h5") | |
| model_dialect = keras.models.load_model("dialect_guessing.h5") | |
| #tokenizer = Tokenizer() | |
| #def prediction(input): | |
| #output_standardform = model_standardform.predict(input) | |
| #output_dialect = model_dialect.predict(input) | |
| #return(output_dialect) | |
| #def predict_dialect(input_word): | |
| # Tokenize the input word | |
| # input_sequence = tokenizer.texts_to_sequences([input_word]) | |
| # if not input_sequence or not input_sequence[0]: | |
| # return "Word not in vocabulary" | |
| # Make a prediction using the model | |
| # predicted_indices = model_dialect.predict(input_sequence) | |
| # Find the index with the highest probability | |
| #predicted_index = predicted_indices.argmax() | |
| # Define a list of dialect labels | |
| #dialect_labels = ["Dialect1", "Dialect2", "Dialect3"] # Customize as needed | |
| # Get the predicted dialect label | |
| #predicted_dialect = dialect_labels[predicted_index] | |
| #return predicted_dialect | |
| def prediction(input_word): | |
| # Tokenize the input word | |
| input_sequence = tokenizer.texts_to_sequences([input_word])[0] | |
| if not input_sequence or not input_sequence[0]: | |
| return "Word not in vocabulary" | |
| # Make a prediction using the model | |
| predicted_indices = model_standardform.predict(np.array(input_sequence)) | |
| # Find the index with the highest probability | |
| predicted_index = np.argmax(predicted_indices) | |
| # Convert the predicted index back to a word using the tokenizer | |
| predicted_word = tokenizer.index_word.get(predicted_index, "Word not in vocabulary") | |
| return predicted_word | |
| desc = "Have you encountered a Finnish dialect or slang word and you cannot find it in a dictionary? This translator predicts the Finnish standard form of the word to enable you to check those words in your usual Finnish sources! Try it out! But, it needs to be stated that not all predictions will be right, although maybe still helpful..." | |
| #def testfunction(input): | |
| # output = input.lower() | |
| # return(output) | |
| demo = gr.Interface(fn=prediction , inputs = gr.Textbox(placeholder="Write a Finnish slang or dialct word here."), outputs = gr.Textbox(placeholder="This is the predicted Finnish standard form of the word."), title = "Local Finnish Translator", description = desc) | |
| demo.launch() | |
| #gr.Textbox(placeholder="This is the predicted location where the word may be used."), |