panda361/Graduate
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

时空ver最后的回忆

Browse Source
This commit is contained in:
panda361
2026-03-19 02:28:50 +08:00
parent 61174433d0
commit cf02f82db0
172 changed files with 22604 additions and 441 deletions
Download Patch File Download Diff File Expand all files Collapse all files

59
0.3_abbrs.tex Normal file
View File

@@ -0,0 +1,59 @@
% 缩写定义
\Abbreviations{ \centering
\renewcommand{\arraystretch}{1.2}
\begin{tabular}{lll}
\textbf{外文缩略字母} & \textbf{外文全称} & \textbf{中文说明}\\
\midrule
% 模型与架构
LLM & Large Language Model & 大语言模型\\
FFN & Feed-Forward Network & 前馈网络\\
GQA & Grouped Query Attention & 分组查询注意力\\
MHSA & Multi-Head Self-Attention & 多头自注意力\\
MoE & Mixture of Experts & 混合专家\\
RoPE & Rotary Position Embedding & 旋转位置编码\\
\\
% 适配方法（通用）
PEFT & Parameter-Efficient Fine-Tuning & 参数高效微调\\
LoRA & Low-Rank Adaptation & 低秩适配\\
DoRA & Weight-Decomposed Low-Rank Adaptation & 权重分解低秩适配\\
AdaLoRA & Adaptive Low-Rank Adaptation & 自适应低秩适配\\
SFT & Supervised Fine-Tuning & 监督微调\\
RAG & Retrieval-Augmented Generation & 检索增强生成\\
\\
% 本文提出的方法
CAM & Contextual Attention Modulation & 上下文注意力调制\\
HyCAM & Hybrid Contextual Attention Modulation & 混合上下文注意力调制\\
RoSA & RoPE-aware Selective Adaptation & RoPE感知选择性适配\\
RoAE & RoPE-aware Attention Enhancement & RoPE感知注意力增强\\
DLS & Dynamic Layer Selection & 动态层选择\\
DyPAM & Dynamic Positional Attention Modulation & 动态位置注意力调制\\
CASCADE & Coarse-to-Fine Spectral Cascading & 从粗到细频谱级联\\
MESSA & Multi-task Efficient Shared-Specific & 多任务高效共享-特有\\
& \quad Sparse Adaptation & \quad 稀疏适配\\
\\
% 信号处理
DCT & Discrete Cosine Transform & 离散余弦变换\\
IDCT & Inverse Discrete Cosine Transform & 逆离散余弦变换\\
FiLM & Feature-wise Linear Modulation & 特征级线性调制\\
\\
% 评测与数据
POI & Point of Interest & 兴趣点\\
QA & Question Answering & 问答\\
ETA & Estimated Time of Arrival & 预计到达时间\\
GPS & Global Positioning System & 全球定位系统\\
WGS84 & World Geodetic System 1984 & 1984世界大地坐标系\\
\\
% 评估指标
MAE & Mean Absolute Error & 平均绝对误差\\
RMSE & Root Mean Square Error & 均方根误差\\
MAPE & Mean Absolute Percentage Error & 平均绝对百分比误差\\
HR & Hit Ratio & 命中率\\
NDCG & Normalized Discounted Cumulative Gain & 归一化折损累计增益\\
BLEU & Bilingual Evaluation Understudy & 双语评估替补\\
\\
% 深度学习基础
CNN & Convolutional Neural Network & 卷积神经网络\\
RNN & Recurrent Neural Network & 循环神经网络\\
GNN & Graph Neural Network & 图神经网络\\
\end{tabular}
}