모델만드는방법.py

Keras 모델 구축 방법 종합 가이드

핵심 개념

• Sequential API
• Functional API (다중 입력/출력)
• Model 서브클래싱
• 사용자 정의 지표 (Custom Metrics)
• 콜백 (EarlyStopping, ModelCheckpoint)
• TensorBoard 시각화

from tensorflow import keras
from tensorflow.keras import layers
from tensorflow.keras.datasets import mnist
from keras.utils import to_categorical

# 1. Sequential API - 기본 방법
model = keras.Sequential([
    layers.Dense(64, activation="relu"),
    layers.Dense(10, activation="softmax")
])
model.build(input_shape=(None, 28*28))

# 2. Sequential API - add 방식
model = keras.Sequential()
model.add(layers.Dense(64, activation="relu"))
model.add(layers.Dense(10, activation="softmax"))
model.build(input_shape=(None, 28*28))

# 3. 입력 크기 미리 지정
model = keras.Sequential()
model.add(keras.Input(shape=(28*28,)))
model.add(layers.Dense(64, activation="relu"))
model.add(layers.Dense(10, activation="softmax"))

# 4. Functional API - 다중 입력/출력
vocabulary_size = 10000
num_tags = 100
num_departments = 4

# 입력 3개
title = keras.Input(shape=(vocabulary_size,), name="title")
text_body = keras.Input(shape=(vocabulary_size,), name="text_body")
tags = keras.Input(shape=(num_tags,), name="tags")

features = layers.Concatenate()([title, text_body, tags])
features = layers.Dense(64, activation="relu")(features)

# 출력 2개
priority = layers.Dense(1, activation="sigmoid", name="priority")(features)
department = layers.Dense(num_departments, activation="softmax", name="department")(features)

model = keras.Model(inputs=[title, text_body, tags],
                    outputs=[priority, department])

# 5. Model 서브클래싱
class CustomerTicketModel(keras.Model):
    def __init__(self, num_departments):
        super().__init__()
        self.concat_layer = layers.Concatenate()
        self.mixing_layer = layers.Dense(64, activation="relu")
        self.priority_scorer = layers.Dense(1, activation="sigmoid")
        self.department_classifier = layers.Dense(num_departments, activation="softmax")

    def call(self, inputs):
        title = inputs["title"]
        text_body = inputs["text_body"]
        tags = inputs["tags"]

        features = self.concat_layer([title, text_body, tags])
        features = self.mixing_layer(features)
        priority = self.priority_scorer(features)
        department = self.department_classifier(features)
        return priority, department

# 콜백 사용하기
callbacks_list = [
    keras.callbacks.EarlyStopping(
        monitor="val_accuracy",
        patience=2,
    ),
    keras.callbacks.ModelCheckpoint(
        filepath="checkpoint_path.keras",
        monitor="val_loss",
        save_best_only=True,
    )
]

# TensorBoard 사용
tensorboard = keras.callbacks.TensorBoard(
    log_dir="./tb_logs",
)
model.fit(train_images, train_labels,
          epochs=10,
          validation_data=(val_images, val_labels),
          callbacks=[tensorboard])

# TensorBoard 실행:
# tensorboard --logdir=./tb_logs --port=6006
# http://localhost:6006/

사용자 정의 훈련 루프

import tensorflow as tf

loss_fn = keras.losses.SparseCategoricalCrossentropy()
optimizer = keras.optimizers.RMSprop()
metrics = [keras.metrics.SparseCategoricalAccuracy()]

def train_step(inputs, targets):
    with tf.GradientTape() as tape:
        predictions = model(inputs, training=True)
        loss = loss_fn(targets, predictions)
    gradients = tape.gradient(loss, model.trainable_weights)
    optimizer.apply_gradients(zip(gradients, model.trainable_weights))
    return loss

# @tf.function으로 성능 향상
@tf.function
def test_step(inputs, targets):
    predictions = model(inputs, training=False)
    loss = loss_fn(targets, predictions)
    return loss