Keras/TensorFlow是否只能处理每批固定大小的输入?

Here

%pylab inline
import random
import string
import numpy as np
import tensorflow as tf
from tensorflow.python.keras.models import Sequential
from tensorflow.python.keras.layers import LSTM, TimeDistributed, Dense, SimpleRNN

BATCH_SIZE = 1
NAMESPACE = string.ascii_uppercase+string.digits
NAMESPACELENGTH = len(NAMESPACE)

def generate_encoding():
    encoding = {}
    for i, letter in enumerate(NAMESPACE):
        encoding[letter] = i
    return encoding

ENCODING = generate_encoding()

def letter_to_vec(letter):
    vec = np.zeros(NAMESPACELENGTH)
    vec[ENCODING[letter]] = 1
    return vec

def word_to_matrix(word):
    return np.array([letter_to_vec(letter) for letter in word])

def dummy_X_y(size=BATCH_SIZE):
    X = []
    y = []
    for N in np.random.randint(1, 100, size):
        tmp = ''.join(random.choice(NAMESPACE) for _ in range(N))
        X.append(tmp)
        y.append(len(tmp))
        del tmp
    return X, y

print(dummy_X_y())


def generate_model():
    model = Sequential()
    model.add(SimpleRNN(1, input_shape=(None, NAMESPACELENGTH), return_sequences=False))
    model.add(Dense(1, activation='linear'))
    model.compile(loss='mean_squared_error', 
                  optimizer='adam', metrics=['mean_squared_error'])
    return model

model = generate_model()

for _ in range(20000):
    # train LSTM
    # generate new random sequence
    X,y = dummy_X_y(size=BATCH_SIZE)
    X = np.array([word_to_matrix(x) for x in X])
    # fit model for one epoch on this sequence
    model.fit(X, y, epochs=1, batch_size=BATCH_SIZE, verbose=1)