tf 2.x keras: Losses,Optimizers,metrics 损失函数优化器度量学习率排程

关注

原创最新推荐文章于 2026-08-27 11:51:22 发布 369 阅读 AI 写同款· GEO 优化 ›

TensorFlow 2.0 十大核心更新：从 eager 执行到 SavedModel 的工程实践指南 TensorFlow 2.0 并非简单版本升级，而是面向工程效率的范式重构。其核心围绕 Eager Execution（即时执行）与 Keras 统一 API 展开，通过默认启用动态图、废除 variable_scope、强化 tf.function 编译机制等设计，显著降低调试成本与协作门槛。技术价值体现在开发可读性提升、训练循环简化、分布式扩展零侵入、模型部署标准化（SavedModel）四大维度。典型应用场景包括推荐系统快速迭代、NLP 模型微调、CV 模型服务化及金融风控实时推理——尤其在需频繁 A 阅读详情

tf 2.x keras

Losses,Optimizers,metrics

https://www.tensorflow.org/api_docs/python/tf/keras/

https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/keras/

https://keras.io/api/

1.损失函数Losses：

https://www.tensorflow.org/api_docs/python/tf/keras/losses/

https://keras.io/api/losses/

CategoricalCrossentropy，Use this crossentropy loss function when there are two or more label classes. We expect labels to be provided in a one_hot representation.

SparseCategoricalCrossentropy，Use this crossentropy loss function when there are two or more label classes. We expect labels to be provided as integers.

BinaryCrossentropy，Use this cross-entropy loss when there are only two label classes (assumed to be 0 and 1).

KLDivergence Computes Kullback-Leibler divergence loss between y_true and y_pred.

MeanAbsoluteError: Computes the mean of absolute difference between labels and predictions.

MeanAbsolutePercentageError: Computes the mean absolute percentage error between y_true and y_pred.

MeanSquaredError: Computes the mean of squares of errors between labels and predictions.

MeanSquaredLogarithmicError: Computes the mean squared logarithmic error between y_true and y_pred.

2.优化器Optimizers：

https://www.tensorflow.org/api_docs/python/tf/keras/optimizers

https://keras.io/api/optimizers/

class Adadelta: Optimizer that implements the Adadelta algorithm.
class Adagrad: Optimizer that implements the Adagrad algorithm.
class Adam: Optimizer that implements the Adam algorithm.
class Adamax: Optimizer that implements the Adamax algorithm.
class Ftrl: Optimizer that implements the FTRL algorithm.
class Nadam: Optimizer that implements the NAdam algorithm.
class RMSprop: Optimizer that implements the RMSprop algorithm.
class SGD: Gradient descent (with momentum) optimizer.
class Optimizer: Base class for Keras optimizers.