# 机器学习与人工智能：模型、训练、评测和常见架构。英文 → 简体中文。
# 只写英文的行保留不译。
Transformer
LLM
GPT
BERT
CNN
RNN
LSTM
GAN
VAE
ViT
GNN
MoE
RAG
RLHF
NLP
SGD
Adam
ReLU
softmax
dropout
token
AUC
machine learning = 机器学习
deep learning = 深度学习
neural network = 神经网络
convolutional neural network = 卷积神经网络
recurrent neural network = 循环神经网络
graph neural network = 图神经网络
generative adversarial network = 生成对抗网络
variational autoencoder = 变分自编码器
autoencoder = 自编码器
diffusion model = 扩散模型
large language model = 大语言模型
foundation model = 基础模型
vision transformer = 视觉 Transformer
mixture of experts = 混合专家
attention mechanism = 注意力机制
self-attention = 自注意力
multi-head attention = 多头注意力
attention head = 注意力头
positional encoding = 位置编码
residual connection = 残差连接
layer normalization = 层归一化
batch normalization = 批归一化
activation function = 激活函数
embedding = 嵌入
encoder = 编码器
decoder = 解码器
latent space = 隐空间
latent variable = 隐变量
tokenizer = 分词器
context window = 上下文窗口
prompt = 提示词
chain-of-thought = 思维链
retrieval-augmented generation = 检索增强生成
zero-shot = 零样本
few-shot = 少样本
pre-training = 预训练
pretraining = 预训练
fine-tuning = 微调
knowledge distillation = 知识蒸馏
quantization = 量化
inference = 推理
hallucination = 幻觉
reinforcement learning = 强化学习
supervised learning = 监督学习
unsupervised learning = 无监督学习
self-supervised learning = 自监督学习
reward model = 奖励模型
policy network = 策略网络
agent = 智能体
training set = 训练集
validation set = 验证集
test set = 测试集
data augmentation = 数据增强
annotation = 标注
ground truth = 真实标签
loss function = 损失函数
objective function = 目标函数
cross-entropy = 交叉熵
gradient descent = 梯度下降
stochastic gradient descent = 随机梯度下降
backpropagation = 反向传播
optimizer = 优化器
learning rate = 学习率
batch size = 批大小
epoch = 轮次
hyperparameter = 超参数
weight decay = 权重衰减
regularization = 正则化
overfitting = 过拟合
underfitting = 欠拟合
generalization = 泛化
convergence = 收敛
classifier = 分类器
feature map = 特征图
feature extraction = 特征提取
accuracy = 准确率
F1 score = F1 分数
confusion matrix = 混淆矩阵
ROC curve = ROC 曲线
perplexity = 困惑度
cross-validation = 交叉验证
ablation study = 消融实验
baseline = 基线
benchmark = 基准
object detection = 目标检测
image classification = 图像分类
semantic segmentation = 语义分割
natural language processing = 自然语言处理
computer vision = 计算机视觉
precision-recall curve = 精确率-召回率曲线
mixed precision = 混合精度
