ValueError: Shape must be rank 2 but is rank 1 for 'MatMul'

Question

I am trying to run a linear regression model using TensorFlow. I have given the code below. However, I got the error as: ValueError: Shape must be at least rank 2 but is rank 1 for 'model_19/MatMul' (op: 'BatchMatMulV2') with input shapes: [1], ?.

From the error, it seems that the input to function model_linear is creating the problem. Any suggestions will be highly appreciating to solve the error.

import tensorflow as tf
x_train = [1.0, 2.0, 3.0, 4.0]
y_train = [1.5, 3.5, 5.5, 7.5]

def model_linear(x, y):
    with tf.variable_scope('model', reuse=tf.AUTO_REUSE):
        W = tf.get_variable("W", initializer=tf.constant([0.1]))
        b = tf.get_variable("b", initializer=tf.constant([0.0]))       
        output = tf.matmul(W, x) + b
        loss = tf.reduce_sum(tf.square(output - y))
    return loss

optimizer = tf.train.GradientDescentOptimizer(0.01)

with tf.Session():
    tf.global_variables_initializer().run()    
    x = tf.placeholder(tf.float32)
    y = tf.placeholder(tf.float32)
    loss = model_linear(x, y)
    train = optimizer.minimize(loss)

    for i in range(1000):
        train.run(feed_dict = {x:x_train, y:y_train})

Answer 1

tf.matmul expects rank 2 tensors, ie, matrices. Instead you have flat vectors. Try x.reshape(-1,1) or x.expand_dims(0) . And it seems that you also need that for your weight matrix.