设计目标

基本架构

工作形态

组件架构

设计原则

1. Dataflow graphs of primitive operators（用原始的op，+，-之类的，而不是复杂的layer操作）
2. Deferred execution (延迟执行，定义和执行分开)
3. Common abstraction for heterogeneous accelerators

示例

# 1. Construct a graph representing the model.
x = tf.placeholder(tf.float32, [BATCH_SIZE, 784]) # Placeholder for input.
y = tf.placeholder(tf.float32, [BATCH_SIZE, 10]) # Placeholder for labels.

W_1 = tf.Variable(tf.random_uniform([784, 100])) # 784x100 weight matrix.
b_1 = tf.Variable(tf.zeros([100])) # 100-element bias vector.
layer_1 = tf.nn.relu(tf.matmul(x, W_1) + b_2) # Output of hidden layer.

W_2 = tf.Variable(tf.random_uniform([100, 10])) # 100x10 weight matrix.
b_2 = tf.Variable(tf.zeros([10])) # 10-element bias vector.
layer_2 = tf.matmul(layer_1, W_2) + b_2 # Output of linear layer.

# 2. Add nodes that represent the optimization algorithm.
loss = tf.nn.softmax_cross_entropy_with_logits(layer_2, y)
train_op = tf.train.AdagradOptimizer(0.01).minimize(loss)

# 3. Execute the graph on batches of input data.
with tf.Session() as sess: # Connect to the TF runtime.
	sess.run(tf.initialize_all_variables()) # Randomly initialize weights.
	for step in range(NUM_STEPS): # Train iteratively for NUM_STEPS.
		x_data, y_data = ... # Load one batch of input data.
		sess.run(train_op, {x: x_data, y: y_data}) # Perform one training step.

Reference

TensorFlow: A System for Large-Scale Machine Learning