概念索引
核心概念 · 术语速查 · 按分类浏览 · 共 494 个概念
高效推理与部署
53 个 动态路由
Acceptance Rate
ACT
adaptive computation
ADEPT
AnchorAttention
CALM
collaborative inference
Cumulative Acceptance Rate
DeeBERT
DSA
dynamic depth
EAGLE
early exit
edge AI
EE-LLM
Element-wise LUT
ELUT
FAST
Fast Graph Decoder
FlashAttention
GGUF
HippoRAG2
KaLMv2
KV-Cache
LADE
LayerSkip
llama.cpp
Medusa
MemGen
MemoryAgent
MoBA
MoD
mpGEMM
Multi-Token Prediction
PagedAttention
PyramidKV
RAG
ReAct
ScissorHands
SIMD
SkipDecode
SnapKV
SpecEE
Speculative Decoding
split computing
StreamingLLM
structural reparameterization
SVD-Softmax
T-MAC
Tool Use
Top-K Selection
Vocabulary Trimming
基础理论
60 个 遍历动力系统
残差能量
测度论
归一化第二矩
核方法
核函数
计算复杂度
模运算
球面码
全变差
时间差分
贪心算法
无放回采样
误差-性能折衷
消融
信息检索
压缩顺序定理
振荡模式
整数线性规划
正交投影
Calibration
L16
Chain-of-Thought
L19
Cholesky 分解
CKA
DCT
Effective Rank
Emergence
Excess Kurtosis
Extended Golay Code
Frobenius Norm
Golay 码
Gram-Schmidt 正交化
Hallucination
L16
In-Context Learning
Lebesgue 测度
Leech lattice
Long-Tailed Distribution
MAD-Sigmoid
Maximal Marginal Relevance
Mechanistic Interpretability
L15
MILP
MMR
Monte Carlo 估计
Multi-Probe LSH
multi-probe retrieval
neural scaling law
NMF
NTK
Rate-Distortion
Rényi 熵
RKHS
Scaling Laws
L16 L19
Shannon 率失真
Soft-OR
Sparse Autoencoder
Spherical K-Means
SVD
Taylor 展开
Test-Time Compute
L19
Tree-of-Thought
剪枝与稀疏化
52 个 幅度剪枝
剪枝压缩比
结构化剪枝
可学习门控
网络剪枝
ART
Block Influence
CDPruner
DeepHoyer
depth pruning
DivPrune
DynamicViT
EfficientVLA
EViT
FastV
GOHSP
GraSP
IMP
LAMP
LayerDrop
LLM-Pruner
LTH
Magnitude Pruning
Minitron
NAP
PACT
PruMerge
PuDDing
RIA
RigL
S2ViT
ShortGPT
SLEB
SliceGPT
SNIP
SparseGPT
SparseVLM
STE
Taylor pruning
TCA
Token Merging
Token Pruning
UPDP
VisionZip
VisPruner
Visual Token Pruning
ViT-Slim
VLA-Cache
Wanda
WDPruning
width pruning
WoodFisher
量化与低秩
52 个 标量量化
低秩分解
混合精度
量化分布
向量量化
AdaLoRA
Adapter
AQLM
ASVD
AutoBit
AWQ
BitNet
BitNet b1.58
DFMC
DoRA
GPTQ
Hadamard rotation
HALO
HGQ
HQQ
InfLoRA
INT4 量化
JointQ
KIVI
LoRA
L18
MXFP
MXINT
NormalFloat
PEFT
L18
PiSSA
Prefix Tuning
Prompt Tuning
PSAQ
PSAQ-ViT
PTQ
PV-tuning
QAT
QJL
QLoRA
QTIP
Quarot
Quip#
RieM
RotorQuant
RVQ
Shape-Gain 量化
SliM-LLM
SmoothQuant
SpinQuant
SpQR
SQNR
SqueezeLLM
模型增长
6 个深度学习基础
57 个 残差连接
多数投票
二分搜索
焦点频率损失
温度缩放
信息熵
形态学操作
余弦相似度
自回归解码
自适应阈值
Activation Function
AdaLN
Backpropagation
Batch Normalization
Binary Cross-Entropy
Computation Graph
Cross-Attention
Cross-Entropy Loss
Cross-Modal Attention
DeepNorm
Dropout
GELU
Gradient Clipping
Gradient Descent
Greedy Decoding
Gumbel-Sigmoid
Gumbel-Softmax
Hadamard 变换
Hadamard 乘积
hierarchical softmax
InfoNCE
Jacobian
Key-Value Projection
KL散度
LayerNorm
Linear Attention
LM Head
Log-Softmax
Mean Pooling
Multi-Head Attention
Multi-Head Self-Attention
Multi-Task Loss
online softmax
Positional Encoding
PreNorm
RMSNorm
Self-Attention
Sigmoid
SiLU
Sobel 边缘检测
Softmax
Softmax Attention
sparse attention
Stop-Gradient
SubLN
Temperature Sampling
Vanishing Gradient
视觉任务
7 个数据集与评估
35 个 BBH
Benchmarking
BoolQ
CIFAR-10
CIFAR-100
COCO
FineWeb-Edu
FLOPs
GenEval
GPQA
GSM8K
HellaSwag
HotpotQA
HumanEval
ImageNet
IoU
LIBERO
LiveCodeBench
LLM Evaluation
LLM-as-Judge
LongBench
LongVideoBench
MMLU
MNIST
MTEB
MuSiQue
NarrativeQA
OpenEQA
perplexity
ScanQA
TruthfulQA
VideoMME
WikiText-103
Wikitext-2
WinoGrande
网络架构
88 个 残差连接
循环神经网络
ALiBi
BeiT
BERT
BigBird
CaiT
classification head
CLIP
ConvGRU
ConvNext
CycleNet
Decision Transformer
DeepSeek
DeepViT
DeiT
DeltaNet
DenseFormer
DenseNet
DINOv2
DINOv3
DiT
DUNE
EfficientVLA
Encoder-Decoder
EVA-CLIP
FLUX
Gemma
GQA
GRU
Highway Network
HunyuanVideo
I-JEPA
ImageBind
Implicit Neural Representation
INR
InternVL3
KAN
Kimi Linear
LayerScale
LLaMA
LLaVA
LLaVA-OneVision
Longformer
LSTM
LVLM
MambaVision
MaskGIT
mHC
Mistral
Mistral-7B
MLP 模块
MoE
NVILA
OpenVLA
PatchTST
Phi-4
PVT
Qwen2.5
Qwen2.5-VL
Qwen3
RADIO
ResNet
ResNet18
ResNet50
ReZero
RNN
RoBERTa
RoPE
RWKV
SDXL
Seq2Seq
SigLIP
SigLIP2
SIREN
SnapGen
SparseTSF
SwinV2
TimeKAN
TimeMixer
TimesNet
Transformer
VAR
VideoMAE
ViT
VLA
VQGAN
WideResNet
训练优化
39 个 AdamW
Alignment
BALD
BatchBALD
Constitutional AI
L16
continual learning
Cosine Annealing
Cosine Decay
CosineAnnealingLR
Curriculum Learning
DDPG
DeepSpeed
DiLoCo
DPO
DreamBooth
EASY
EMA
EWC
FedAvg
Fine-tuning
L18
GaLore
GRPO
L19
Hessian
L1 正则化
Muon
pipeline parallelism
PPO
Pretraining
Reward Model
Ridge Regression
RLHF
L15 L16
SAM
SFT
SOAP
Soft Actor-Critic
Supervised Contrastive Loss
Synthetic Data
L16 L19
ZeRO
μP