app.py

import numpy as np
import tensorflow as tf
from huggingface_hub import from_pretrained_keras

import gradio as gr

## -- DIGIT ENCODER AND DECODER -- ##
class CharacterTable:
    """Given a set of characters:
    + Encode them to a one-hot integer representation
    + Decode the one-hot or integer representation to their character output
    + Decode a vector of probabilities to their character output
    """

    def __init__(self, chars):
        """Initialize character table.
        # Arguments
            chars: Characters that can appear in the input.
        """
        self.chars = sorted(set(chars))
        self.char_indices = dict((c, i) for i, c in enumerate(self.chars))
        self.indices_char = dict((i, c) for i, c in enumerate(self.chars))

    def encode(self, C, num_rows):
        """One-hot encode given string C.
        # Arguments
            C: string, to be encoded.
            num_rows: Number of rows in the returned one-hot encoding. This is
                used to keep the # of rows for each data the same.
        """
        x = np.zeros((num_rows, len(self.chars)))
        for i, c in enumerate(C):
            x[i, self.char_indices[c]] = 1
        return x

    def decode(self, x, calc_argmax=True):
        """Decode the given vector or 2D array to their character output.
        # Arguments
            x: A vector or a 2D array of probabilities or one-hot representations;
                or a vector of character indices (used with `calc_argmax=False`).
            calc_argmax: Whether to find the character index with maximum
                probability, defaults to `True`.
        """
        if calc_argmax:
            x = x.argmax(axis=-1)
        return "".join(self.indices_char[x] for x in x)
        
   
   
## -- INFERENCE CODE -- ##
def check_num_validity(a, b):
    """Validates the inputs before feeding to the model.
       Checks if both the inputs are positive integers and are not more than 5 digits.
    """
    if a.isdigit() and b.isdigit():
        if len(a) <= 5 and len(b) <= 5:
            return True, None
        else:
            return False, "Input can contain maximum of five digits"
    else:
        return False, "Input has to be positive integers (each can be max 5 digits)"

def add_2_nums(a, b):
    """Performs the model inference after input validation.
    """
    a = a.strip()
    b = b.strip()
    
    ip_val_op = check_num_validity(a, b)

    if ip_val_op[0]:
        # Input encoding
        q = f"{a.strip()}+{b.strip()}"
        query = q + " " * (MAXLEN - len(q))
        rev_query = query[::-1]
        
        inp = ctable.encode(rev_query, MAXLEN)
        inp = np.expand_dims(inp, axis=0)

        # Prediction and output decoding
        preds = np.argmax(model.predict(inp), axis=-1)
        guess = ctable.decode(preds[0], calc_argmax=False)

        return guess, rev_query, query
    else:
        return "", "Error", ip_val_op[1]
   
   
## -- LSTM INFERENCE SETUP --##
# Setup for encoding and decoding the inputs     
chars = "0123456789+ "
ctable = CharacterTable(chars)
DIGITS = 5
REVERSE = True

# Maximum length of input is 'int + int' (e.g., '345+678'). Maximum length of
# int is DIGITS.
MAXLEN = DIGITS + 1 + DIGITS

# Model
model = from_pretrained_keras("vdprabhu/addition-lstm")

## -- GRADIO SETUP -- ##
inputs = [gr.Textbox(), gr.Textbox()]
outputs = [gr.Textbox(label="LSTM Output"), gr.Textbox(label="LSTM Input"), gr.Textbox(label="User Query")]

examples = [[53511,98888], [452,12]] 
title = "Addition using LSTM"

more_text = "It is interesting to note that the input is reversed before feeding to LSTM. Sequence order inversion introduces shorter term dependencies between source and target for this problem."
description = f"LSTM model is used to add two numbers, provided as strings.\n\n{more_text}"

article = """
<p style='text-align: center'>
    <a href='https://keras.io/examples/nlp/addition_rnn/' target='_blank'>Keras Example by Smerity and others</a>
    <br>
    Space by Vrinda Prabhu
    </p>
"""

gr.Interface(fn=add_2_nums, inputs=inputs, outputs=outputs,  examples=examples, article=article, allow_flagging="never", analytics_enabled=False,
  title=title, description=description).launch(enable_queue=True)