Глоссарий
Ключевые термины искусственного интеллекта — коротко и по делу.
А
- A2C A2C
- A3C A3C
- Ablation study Ablation study
- Absolute positional encoding Absolute positional encoding
- Abstractive summarization Abstractive summarization
- A/B testing A/B testing
- Accountability Accountability
- Accuracy Accuracy
- Action recognition Action recognition
- Action space Action space
- Activation function Activation function
- Active inference serving Active inference serving
- Active learning loop Active learning loop
- Actor-Critic Actor-Critic
- AdaGrad AdaGrad
- Adam Adam
- AdamW AdamW
- Adapter Adapter
- Advantage Advantage
- Adversarial attack Adversarial attack
- Adversarial defense Adversarial defense
- Adversarial example Adversarial example
- Adversarial training Adversarial training
- Agentic RAG Agentic RAG
- AgentLess AgentLess
- Agent memory store Agent memory store
- Agent trajectory Agent trajectory
- AI Act AI Act
- AI literacy AI literacy
- AI Safety AI Safety
- ALBERT ALBERT
- Albumentations Albumentations
- ALiBi ALiBi
- ALIGN ALIGN
- Alignment Alignment
- Alignment problem Alignment problem
- All-reduce All-reduce
- AlpacaEval AlpacaEval
- AlphaFold AlphaFold
- AlphaGo AlphaGo
- AlphaZero AlphaZero
- AMP AMP
- Anchor box Anchor box
- ANN index ANN index
- ANN search ANN search
- Anomaly detection in time series Anomaly detection in time series
- Answer relevance Answer relevance
- Answer-then-verify Answer-then-verify
- APE APE
- API synthesis API synthesis
- API tool API tool
- ARC ARC
- Arena Elo Arena Elo
- Arena hard Arena hard
- ARIMA ARIMA
- Artifact store Artifact store
- ASR ASR
- Assistant message Assistant message
- AST AST
- Async inference Async inference
- Attention Attention
- Attention sink Attention sink
- Attention transfer Attention transfer
- Attribution Attribution
- AUC AUC
- Audio captioning Audio captioning
- Audio tokenization Audio tokenization
- Autocomplete Autocomplete
- Automatic prompt engineer Automatic prompt engineer
- Autonomous weapon Autonomous weapon
- Auxiliary loss Auxiliary loss
- Avatar Avatar
- Average pooling Average pooling
- Автономный агент Автономный агент
- Автоэнкодер Автоэнкодер
- Агент Агент
- Активное обучение Активное обучение
- Аномалия Аномалия
- Анонимизация Анонимизация
- Ансамбль Ансамбль
Б
- Backbone Backbone
- Backdoor attack Backdoor attack
- Backtesting Backtesting
- Balanced dataset Balanced dataset
- Bandit Bandit
- BART BART
- Baseline model Baseline model
- Batch inference Batch inference
- Batch size Batch size
- Bayesian statistics Bayesian statistics
- Beam search Beam search
- Behavioral cloning Behavioral cloning
- BEIR BEIR
- BERT BERT
- BERTScore BERTScore
- BF16 BF16
- Bias Bias
- Bias-variance tradeoff Bias-variance tradeoff
- Bias в данных Bias в данных
- Bidirectional LSTM Bidirectional LSTM
- Bi-encoder Bi-encoder
- BIG-bench BIG-bench
- BigBird BigBird
- Biosecurity risk Biosecurity risk
- Biosignal ML Biosignal ML
- BitFit BitFit
- Blackboard architecture Blackboard architecture
- BLEU BLEU
- BLIP BLIP
- BLOOM BLOOM
- BM25 BM25
- Boilerplate generation Boilerplate generation
- BoolQ BoolQ
- Bootstrap Bootstrap
- BOS token BOS token
- Bottleneck Bottleneck
- Bounding box Bounding box
- Brier score Brier score
- Browser agent Browser agent
- Bug localization Bug localization
- BYOL BYOL
- Byte-level BPE Byte-level BPE
- Байесовский вывод Байесовский вывод
- Батч-нормализация Батч-нормализация
- Бенчмарк Бенчмарк
- Бернулли Бернулли
- Блендинг Блендинг
- Бустинг Бустинг
- Бэггинг Бэггинг
- Бэтч Бэтч
В
- Question answering Question answering
- VAD VAD
- VAE VAE
- Value function Value function
- Vanishing activations Vanishing activations
- Verifier Verifier
- Verifier model Verifier model
- VGG VGG
- Video understanding Video understanding
- Virtual adversarial training Virtual adversarial training
- Vision-language-action Vision-language-action
- Vision-language model Vision-language model
- Visual question answering Visual question answering
- ViT ViT
- VLA VLA
- vLLM vLLM
- Voice activity detection Voice activity detection
- Voice cloning Voice cloning
- Voice conversion Voice conversion
- VQA VQA
- Wake word Wake word
- Walk-forward validation Walk-forward validation
- Warmup Warmup
- Wasserstein distance Wasserstein distance
- Watermarking Watermarking
- Wav2Vec Wav2Vec
- WaveNet WaveNet
- W&B W&B
- Weak supervision Weak supervision
- Weaviate Weaviate
- Weight decay Weight decay
- Weight initialization Weight initialization
- Weight streaming Weight streaming
- Whisper Whisper
- WinoGrande WinoGrande
- WiSE-FT WiSE-FT
- WordPiece WordPiece
- Word sense disambiguation Word sense disambiguation
- Workflow Workflow
- World model World model
- World model for robots World model for robots
- Валидация Валидация
- Вектор Вектор
- Векторная БД Векторная БД
- Выпуклая функция Выпуклая функция
Г
- GAN GAN
- GAT GAT
- Gateway Gateway
- Gaussian Mixture Gaussian Mixture
- Gaussian Splatting Gaussian Splatting
- GCN GCN
- GDPR и AI GDPR и AI
- GELU GELU
- Gemini Gemini
- Gemma Gemma
- Generate-then-retrieve Generate-then-retrieve
- Generative QA Generative QA
- Generator Generator
- Genomics foundation model Genomics foundation model
- GGUF GGUF
- Global average pooling Global average pooling
- Golden set Golden set
- Gold standard Gold standard
- Gopher Gopher
- GPT GPT
- GPT-2 GPT-2
- GPT-3 GPT-3
- GPT-4 GPT-4
- GPU memory bandwidth GPU memory bandwidth
- GPU utilization GPU utilization
- GQA GQA
- Gradient accumulation Gradient accumulation
- Gradient clipping Gradient clipping
- Gradient noise scale Gradient noise scale
- Grammar sampling Grammar sampling
- Grapheme Grapheme
- Graph Neural Network Graph Neural Network
- Graph of Thoughts Graph of Thoughts
- Graph optimization Graph optimization
- GraphRAG GraphRAG
- Grasp detection Grasp detection
- Great Expectations Great Expectations
- Greedy decoding Greedy decoding
- Groundedness Groundedness
- Grounding Grounding
- Grounding DINO Grounding DINO
- Grouped-Query Attention Grouped-Query Attention
- Group Normalization Group Normalization
- Group split Group split
- GRU GRU
- GSM8K GSM8K
- Guardrails Guardrails
- GUI agent GUI agent
- Guideline Guideline
- Галлюцинация Галлюцинация
- Генеративная модель Генеративная модель
- Гессиан Гессиан
- Гиперпараметр Гиперпараметр
- Глобальный минимум Глобальный минимум
- Градиент Градиент
- Градиентный бустинг Градиентный бустинг
- Градиент функции Градиент функции
Д
- DAG агентов DAG агентов
- DALL·E DALL·E
- Dangerous capability Dangerous capability
- Data curriculum Data curriculum
- Data drift Data drift
- Data-flow analysis Data-flow analysis
- Data flywheel Data flywheel
- Data-free distillation Data-free distillation
- Data leakage Data leakage
- Data parallelism Data parallelism
- Data poisoning Data poisoning
- Data quality Data quality
- Datasheet Datasheet
- Data versioning Data versioning
- DBSCAN DBSCAN
- DDPG DDPG
- Dead ReLU Dead ReLU
- Debate Debate
- DeBERTa DeBERTa
- Deceptive alignment Deceptive alignment
- Decode phase Decode phase
- Deduplication Deduplication
- Deepfake Deepfake
- DeepSeek DeepSeek
- DeepSpeed DeepSpeed
- Deep supervision Deep supervision
- Delegation Delegation
- Delimiter prompting Delimiter prompting
- Demographic parity Demographic parity
- DenseNet-блок DenseNet-блок
- Dense retrieval Dense retrieval
- Dense reward Dense reward
- Dependency parsing Dependency parsing
- Depth estimation Depth estimation
- Depthwise convolution Depthwise convolution
- Detection head Detection head
- Determinism Determinism
- DETR DETR
- Devin-style agent Devin-style agent
- Dialogue system Dialogue system
- Diff pruning Diff pruning
- Diff understanding Diff understanding
- Diffusion Transformer Diffusion Transformer
- Digital pathology Digital pathology
- Dilated convolution Dilated convolution
- Dimensionality reduction Dimensionality reduction
- Discount factor Discount factor
- Discrete prompts Discrete prompts
- Discriminator Discriminator
- Disk offload Disk offload
- Disparate impact Disparate impact
- Distant supervision Distant supervision
- DistilBERT DistilBERT
- Distillation temperature Distillation temperature
- Distilled dataset Distilled dataset
- Distributed training Distributed training
- DiT DiT
- Diversity Distinct-n Diversity Distinct-n
- Diversity ranking Diversity ranking
- Document AI Document AI
- DocVQA DocVQA
- Domain adaptation Domain adaptation
- Domain randomization Domain randomization
- Double DQN Double DQN
- DPO DPO
- DQN DQN
- Draft model Draft model
- Drift detection Drift detection
- DROP DROP
- Dropout Dropout
- DropPath DropPath
- Drug discovery AI Drug discovery AI
- DSL generation DSL generation
- DSPy DSPy
- Dual-use Dual-use
- Dueling DQN Dueling DQN
- DVC DVC
- Датасет Датасет
- Действие Действие
- Декодер Декодер
- Дерево решений Дерево решений
- Детекция объектов Детекция объектов
- Дисбаланс классов Дисбаланс классов
- Дистилляция Дистилляция
- Дифференциальная приватность Дифференциальная приватность
- Диффузия Диффузия
Е
- Early fusion Early fusion
- Early stopping Early stopping
- EEG decoding EEG decoding
- EfficientNet EfficientNet
- EHR modeling EHR modeling
- Eigenvalue Eigenvalue
- Eigenvector Eigenvector
- Elasticsearch vector Elasticsearch vector
- Elastic Weight Consolidation Elastic Weight Consolidation
- ELECTRA ELECTRA
- ELU ELU
- EMA weights EMA weights
- Embedding space arithmetic Embedding space arithmetic
- Embedding для изображений Embedding для изображений
- Embedding слой Embedding слой
- Embodied AI Embodied AI
- EnCodec EnCodec
- Ensemble serving Ensemble serving
- Entity embeddings Entity embeddings
- Entity linking Entity linking
- Entropy Entropy
- Environment Environment
- Environment API Environment API
- EOS token EOS token
- Episode Episode
- Episodic memory Episodic memory
- Epsilon-greedy Epsilon-greedy
- Equalized odds Equalized odds
- Ergodicity Ergodicity
- Error analysis Error analysis
- Eval-driven development Eval-driven development
- Eval harness Eval harness
- Eval для агентов Eval для агентов
- Event-driven agent Event-driven agent
- Exact Match Exact Match
- Exact Match@k Exact Match@k
- Executor Executor
- Exemplar ordering Exemplar ordering
- Expectation-Maximization Expectation-Maximization
- Experience replay Experience replay
- Experiment tracking Experiment tracking
- Expert parallelism Expert parallelism
- Explainable AI Explainable AI
- Explain-then-answer Explain-then-answer
- Explicit feedback Explicit feedback
- Exploding gradient Exploding gradient
- Exploitation Exploitation
- Exploration Exploration
- Exposure bias Exposure bias
- Евклидово расстояние Евклидово расстояние
З
И
- IA³ IA³
- IA³ IA³
- ICA ICA
- IFEval IFEval
- Image augmentation Image augmentation
- Image captioning Image captioning
- Image inpainting Image inpainting
- Imagen Imagen
- Image-to-image Image-to-image
- Imitation from video Imitation from video
- Imitation learning Imitation learning
- Impala Impala
- Implicit CTR Implicit CTR
- Implicit feedback Implicit feedback
- Imputation Imputation
- Inception Inception
- In-context learning In-context learning
- Indirect prompt injection Indirect prompt injection
- Inference cost Inference cost
- InfoNCE InfoNCE
- Information gain Information gain
- Information retrieval Information retrieval
- Instance Normalization Instance Normalization
- Instance segmentation Instance segmentation
- InstructGPT InstructGPT
- Instruction tuning Instruction tuning
- INT4 INT4
- INT8 INT8
- Intent classification Intent classification
- Inter-annotator agreement Inter-annotator agreement
- Interpretability Interpretability
- Inverse RL Inverse RL
- IoU IoU
- Isolation Forest Isolation Forest
- IS score IS score
- Iterative magnitude pruning Iterative magnitude pruning
- IVF IVF
- Инференс Инференс
- Итерация Итерация
Й
К
- ColBERT ColBERT
- Coreference resolution Coreference resolution
- Corrective RAG Corrective RAG
- Cross-entropy Cross-entropy
- Kernel Kernel
- Kernel fusion Kernel fusion
- Kernel trick Kernel trick
- Keypoint detection Keypoint detection
- Keyword spotting Keyword spotting
- KILT KILT
- KL-дивергенция KL-дивергенция
- K-means K-means
- k-NN k-NN
- KNN prompting KNN prompting
- Knowledge distillation online Knowledge distillation online
- Knowledge distillation teacher Knowledge distillation teacher
- Knowledge graph Knowledge graph
- KTO KTO
- KV-cache KV-cache
- KV cache quantization KV cache quantization
- Qasper Qasper
- Qdrant Qdrant
- Q-function Q-function
- Q-learning Q-learning
- QLoRA QLoRA
- QR decomposition QR decomposition
- Quadruped locomotion Quadruped locomotion
- Quantization aware training Quantization aware training
- Query expansion Query expansion
- Query rewriting Query rewriting
- Query understanding Query understanding
- Qwen Qwen
- XAI XAI
- Xavier initialization Xavier initialization
- XLA XLA
- XOR-Retrieve XOR-Retrieve
- Категориальное распределение Категориальное распределение
- Квантизация Квантизация
- Классификация Классификация
- Классификация изображений Классификация изображений
- Кластеризация Кластеризация
- Контекст Контекст
- Контекстное окно Контекстное окно
- Корпус Корпус
- Косинусное сходство Косинусное сходство
- Кросс-валидация Кросс-валидация
Л
- L1-регуляризация L1-регуляризация
- L2-регуляризация L2-регуляризация
- Label drift Label drift
- Label encoding Label encoding
- Label smoothing Label smoothing
- Lag features Lag features
- LakeFS LakeFS
- Landmark detection Landmark detection
- Language model Language model
- Late fusion Late fusion
- Latency Latency
- Latency-aware agent Latency-aware agent
- Latent variable Latent variable
- Law of large numbers Law of large numbers
- LayerDrop LayerDrop
- Layer Normalization Layer Normalization
- Layout analysis Layout analysis
- LDA LDA
- Leaky ReLU Leaky ReLU
- Learning rate Learning rate
- Learning rate schedule Learning rate schedule
- Learning to rank Learning to rank
- Least squares Least squares
- Least-to-most prompting Least-to-most prompting
- LegalBench LegalBench
- Lemmatization Lemmatization
- LightGBM LightGBM
- Likelihood Likelihood
- LIME LIME
- Lineage Lineage
- Linear probe Linear probe
- Linformer Linformer
- Lion optimizer Lion optimizer
- Lip sync Lip sync
- Listwise ranking Listwise ranking
- LiveCodeBench LiveCodeBench
- LLaMA LLaMA
- llama.cpp llama.cpp
- LLaVA LLaVA
- LLM LLM
- LLM-as-a-judge LLM-as-a-judge
- LLMOps LLMOps
- LMSYS LMSYS
- Log-loss Log-loss
- Long context Long context
- Longformer Longformer
- Long-tail distribution Long-tail distribution
- Long-term memory Long-term memory
- Lookahead decoding Lookahead decoding
- LoRA LoRA
- LoRA для диффузии LoRA для диффузии
- Loss landscape Loss landscape
- Lost in the middle Lost in the middle
- Lottery Ticket Hypothesis Lottery Ticket Hypothesis
- Low-rank adaptation Low-rank adaptation
- LP-FT LP-FT
- LSTM LSTM
- Лагранжиан Лагранжиан
- Латентное пространство Латентное пространство
- Линейная регрессия Линейная регрессия
- Логистическая регрессия Логистическая регрессия
- Локальный минимум Локальный минимум
- Лосс-функция Лосс-функция
М
- Machine translation Machine translation
- MAE MAE
- Majority voting Majority voting
- Mamba Mamba
- MAML MAML
- Manipulation Manipulation
- Many-shot Many-shot
- mAP mAP
- MAPE MAPE
- Map-free navigation Map-free navigation
- MAP ranking MAP ranking
- MAP-оценка MAP-оценка
- Margin ranking loss Margin ranking loss
- Markov chain Markov chain
- Masked language modeling Masked language modeling
- Mask R-CNN Mask R-CNN
- Mask token Mask token
- Material discovery Material discovery
- MAUVE MAUVE
- Maximum Likelihood Maximum Likelihood
- Max pooling Max pooling
- MBPP MBPP
- MCP MCP
- MCTS MCTS
- MDP MDP
- Mean Teacher Mean Teacher
- Mechanistic interpretability Mechanistic interpretability
- Medical imaging AI Medical imaging AI
- MedQA MedQA
- Medusa Medusa
- Mel-spectrogram Mel-spectrogram
- Membership inference Membership inference
- Memory Memory
- Memory network Memory network
- Mesa-optimization Mesa-optimization
- Message passing Message passing
- Meta-learning Meta-learning
- Meta-prompt Meta-prompt
- METEOR METEOR
- MFCC MFCC
- Midjourney Midjourney
- Milvus Milvus
- MinHash MinHash
- Mini-batch Mini-batch
- MIRACL MIRACL
- Misinformation Misinformation
- Mistral Mistral
- Mixed precision Mixed precision
- MixFinetune MixFinetune
- Mixtral Mixtral
- MixUp MixUp
- MLE MLE
- MLflow MLflow
- MLIR MLIR
- MLOps MLOps
- MLP MLP
- MMLU MMLU
- MMR MMR
- MobileNet MobileNet
- MoCo MoCo
- Mode collapse Mode collapse
- Model-based RL Model-based RL
- Model card Model card
- Model cascade Model cascade
- Model-free RL Model-free RL
- Model inversion Model inversion
- Model monitoring Model monitoring
- Model parallelism Model parallelism
- Model registry Model registry
- Model routing Model routing
- Model server Model server
- Model soup Model soup
- Model stealing Model stealing
- MoE MoE
- Molecular generation Molecular generation
- Momentum Momentum
- Monte Carlo Tree Search Monte Carlo Tree Search
- Mosaic augmentation Mosaic augmentation
- Motion planning Motion planning
- Movement pruning Movement pruning
- MRR MRR
- MRR@10 MRR@10
- MSE MSE
- MS-MARCO MS-MARCO
- MT-Bench MT-Bench
- MTEB MTEB
- Multi-agent RL Multi-agent RL
- Multicollinearity Multicollinearity
- Multi-GPU inference Multi-GPU inference
- Multimodal fusion Multimodal fusion
- Multi-objective ranking Multi-objective ranking
- Multi-object tracking Multi-object tracking
- Multi-Query Attention Multi-Query Attention
- Multi-query retrieval Multi-query retrieval
- MultiRC MultiRC
- Multi-scale training Multi-scale training
- Multi-task learning Multi-task learning
- Multivariate series Multivariate series
- Music generation Music generation
- Mutation testing Mutation testing
- Mutual information Mutual information
- MuZero MuZero
- Манхэттенское расстояние Манхэттенское расстояние
- Марковский процесс решений Марковский процесс решений
- Матрица Матрица
- Матрица ошибок Матрица ошибок
- Метрика Метрика
- Мультиагентная система Мультиагентная система
- Мультимодальность Мультимодальность
Н
- Named Entity Recognition Named Entity Recognition
- NAS NAS
- Natural Questions Natural Questions
- Navigation policy Navigation policy
- N-BEATS N-BEATS
- NCCL NCCL
- NDCG NDCG
- nDCG@10 nDCG@10
- Near-duplicate Near-duplicate
- Neck Neck
- Needle in a haystack Needle in a haystack
- NER NER
- NeRF NeRF
- Nesterov Nesterov
- Neural architecture search Neural architecture search
- Neural Turing Machine Neural Turing Machine
- Next token prediction Next token prediction
- N-gram N-gram
- NLG NLG
- NLP NLP
- NLU NLU
- NMS NMS
- NMT NMT
- NoCaps NoCaps
- Noisy student Noisy student
- Non-maximum suppression Non-maximum suppression
- Novelty detection Novelty detection
- NPU NPU
- NT-Xent NT-Xent
- Nucleus sampling Nucleus sampling
- NVMe KV cache NVMe KV cache
- Награда Награда
- Наивный байес Наивный байес
- Недообучение Недообучение
- Нейросеть Нейросеть
- Норма Норма
- Нормализация Нормализация
- Нормальное распределение Нормальное распределение
О
- Object tracking Object tracking
- Observability Observability
- Observation Observation
- OCR OCR
- OCR pipeline OCR pipeline
- Offline evaluation Offline evaluation
- Offline features Offline features
- Offline RL Offline RL
- Offloading Offloading
- Off-policy Off-policy
- OK-VQA OK-VQA
- One-cycle policy One-cycle policy
- One-hot encoding One-hot encoding
- One-shot One-shot
- Online evaluation Online evaluation
- Online features Online features
- Online inference Online inference
- ONNX ONNX
- ONNX Runtime ONNX Runtime
- On-policy On-policy
- Ontology Ontology
- Open-domain QA Open-domain QA
- OpenHands OpenHands
- OpenTelemetry OpenTelemetry
- OpenVINO OpenVINO
- Open-vocabulary detection Open-vocabulary detection
- Operator fusion Operator fusion
- OPRO OPRO
- OPT OPT
- Optical flow Optical flow
- Orchestrator Orchestrator
- ORPO ORPO
- Orthogonalization Orthogonalization
- Outlier Outlier
- Out-of-bag Out-of-bag
- Output watermark Output watermark
- Overlapping chunks Overlapping chunks
- Оversampling Оversampling
- Обратное распространение Обратное распространение
- Обучение без учителя Обучение без учителя
- Обучение с учителем Обучение с учителем
- Оверфиттинг Оверфиттинг
- Оптимизатор Оптимизатор
- Оптимизация Оптимизация
П
- Padding Padding
- PAD token PAD token
- PagedAttention PagedAttention
- Pairwise ranking Pairwise ranking
- PaLM PaLM
- PAL prompting PAL prompting
- Panoptic segmentation Panoptic segmentation
- Parent document retriever Parent document retriever
- Partial observability Partial observability
- Pass@1 Pass@1
- Pass@k Pass@k
- PCA PCA
- PCA math PCA math
- PEFT PEFT
- PEFT PEFT
- Performer Performer
- Perplexity benchmark Perplexity benchmark
- Personalization Personalization
- Personal search Personal search
- Persona prompting Persona prompting
- Phi Phi
- Phoneme Phoneme
- PickScore PickScore
- PII PII
- Pipeline Pipeline
- Pipeline parallelism Pipeline parallelism
- PIQA PIQA
- Pitch Pitch
- Plan-and-Execute Plan-and-Execute
- Planner Planner
- Plugin Plugin
- Pointer network Pointer network
- Point-in-time join Point-in-time join
- Pointwise convolution Pointwise convolution
- Pointwise ranking Pointwise ranking
- Policy Policy
- Policy gradient Policy gradient
- Polyak averaging Polyak averaging
- POMDP POMDP
- Pooling Pooling
- Pose estimation Pose estimation
- Position bias Position bias
- POS-tagging POS-tagging
- Posterior Posterior
- Post-training quantization Post-training quantization
- Power efficiency Power efficiency
- PPO PPO
- PR agent PR agent
- Precision Precision
- Precision@k Precision@k
- Precision и Recall Precision и Recall
- Preference data Preference data
- Prefill Prefill
- Prefix tuning Prefix tuning
- Presence penalty Presence penalty
- Prior Prior
- Prioritized experience replay Prioritized experience replay
- Privacy Privacy
- Product quantization Product quantization
- Program-aided language model Program-aided language model
- Program repair Program repair
- Progressive growing Progressive growing
- Progressive resizing Progressive resizing
- Prompt cache Prompt cache
- Prompt chaining Prompt chaining
- Prompt injection Prompt injection
- Prompt library Prompt library
- Prompt optimization Prompt optimization
- Prompt tuning Prompt tuning
- Prompt versioning Prompt versioning
- Property-based testing Property-based testing
- Prophet Prophet
- Protein language model Protein language model
- Pruning Pruning
- Pseudo-labeling Pseudo-labeling
- P-Tuning P-Tuning
- PubMedQA PubMedQA
- p-value p-value
- Параметр Параметр
- Перплексия Перплексия
- Перцептрон Перцептрон
- Полуобучение Полуобучение
- Признак Признак
- Производная Производная
- Промпт Промпт
- Промпт-инжиниринг Промпт-инжиниринг
- Псевдонимизация Псевдонимизация
Р
- R² R²
- R2D2 R2D2
- Radiomics Radiomics
- RAG RAG
- RAG по коду RAG по коду
- Rainbow Rainbow
- RAPTOR RAPTOR
- Rate limit Rate limit
- R-Drop R-Drop
- ReAct ReAct
- Realtime inference Realtime inference
- Recall Recall
- Recall@k Recall@k
- Receptive field Receptive field
- Recommendation system Recommendation system
- Recursive reward modeling Recursive reward modeling
- Red teaming Red teaming
- RefCOCO RefCOCO
- Reflection Reflection
- Refusal Refusal
- Regression suite Regression suite
- REINFORCE REINFORCE
- Reinforcement Learning Reinforcement Learning
- Reject option Reject option
- Relation distillation Relation distillation
- Relative positional encoding Relative positional encoding
- ReLU ReLU
- Render-and-train Render-and-train
- Repetition penalty Repetition penalty
- Replay buffer Replay buffer
- Repo agent Repo agent
- Repository-level context Repository-level context
- Reproducibility Reproducibility
- Reranker Reranker
- Residual connection Residual connection
- ResNet ResNet
- Resolution curriculum Resolution curriculum
- Responsible AI Responsible AI
- RetNet RetNet
- Retrieval-augmented generation Retrieval-augmented generation
- Retrieval before generation Retrieval before generation
- Return Return
- Reward hacking Reward hacking
- Reward model Reward model
- Reward shaping Reward shaping
- RLHF RLHF
- RMSE RMSE
- RMSNorm RMSNorm
- RMSProp RMSProp
- RNN RNN
- RoBERTa RoBERTa
- Robot foundation model Robot foundation model
- Robust fine-tuning Robust fine-tuning
- Robustness Robustness
- ROCm ROCm
- ROC-кривая ROC-кривая
- Role prompting Role prompting
- Rolling window Rolling window
- Rollout Rollout
- RoPE RoPE
- Rotary Position Embedding Rotary Position Embedding
- ROUGE ROUGE
- Router agent Router agent
- Router + specialists Router + specialists
- Rubric Rubric
- Rules + ML Rules + ML
- Runtime error rate Runtime error rate
- RWKV RWKV
- Разметка Разметка
- Ранжирование Ранжирование
- Распределение Распределение
- Регрессия Регрессия
- Регуляризация Регуляризация
- Регуляризованная цель Регуляризованная цель
С
- Caching ответов Caching ответов
- Calibration Calibration
- Calibration of ranker Calibration of ranker
- Canary deployment Canary deployment
- Candidate generation Candidate generation
- Canonicalization Canonicalization
- Capability evaluation Capability evaluation
- Capsule network Capsule network
- Carbon cost of training Carbon cost of training
- Catastrophic forgetting Catastrophic forgetting
- CatBoost CatBoost
- CatBoost ranking CatBoost ranking
- Causal attention Causal attention
- Causal inference in medicine Causal inference in medicine
- Central limit theorem Central limit theorem
- Champion-challenger Champion-challenger
- Change point detection Change point detection
- Channel Channel
- ChartQA ChartQA
- ChatGPT ChatGPT
- Chat template Chat template
- Checkpoint averaging Checkpoint averaging
- Checkpointing activation Checkpointing activation
- Chinchilla Chinchilla
- CI/CD для ML CI/CD для ML
- Citation Citation
- Cite-then-explain Cite-then-explain
- Class weight Class weight
- Claude Claude
- Click model Click model
- Climate modeling ML Climate modeling ML
- Clinical NLP Clinical NLP
- CLIP CLIP
- CLIPScore CLIPScore
- Closed-book QA Closed-book QA
- CLS token CLS token
- COCO caption COCO caption
- Codec model Codec model
- Code completion Code completion
- Code interpreter Code interpreter
- Code LLM Code LLM
- Code review bot Code review bot
- Codex Codex
- Cohen's kappa Cohen's kappa
- Coherence score Coherence score
- Cold start Cold start
- Cold-start item Cold-start item
- Cold-start user Cold-start user
- Collaborative filtering Collaborative filtering
- Compilation for inference Compilation for inference
- Compile rate Compile rate
- Compilergpt Compilergpt
- Computer use Computer use
- Concept drift Concept drift
- Confidence interval Confidence interval
- Confidence threshold Confidence threshold
- Config as code Config as code
- Confusion matrix Confusion matrix
- Consistency model Consistency model
- Consistency regularization Consistency regularization
- Constitutional AI Constitutional AI
- Constrained decoding Constrained decoding
- Content-based filtering Content-based filtering
- Content filter Content filter
- Context precision Context precision
- Context recall Context recall
- Context stuffing Context stuffing
- Contextual bandit Contextual bandit
- Continual learning Continual learning
- Continuous batching Continuous batching
- Continuous batching fairness Continuous batching fairness
- Continuous prompts Continuous prompts
- Continuous training Continuous training
- Contract net Contract net
- Contrastive learning Contrastive learning
- Contrastive loss Contrastive loss
- Control-flow graph Control-flow graph
- ControlNet ControlNet
- ConvNeXt ConvNeXt
- Convolution math Convolution math
- Copilot Copilot
- Core ML Core ML
- Correlation Correlation
- Correlation vs causation Correlation vs causation
- Cosine annealing Cosine annealing
- Cost-aware agent Cost-aware agent
- Cost/quality tradeoff Cost/quality tradeoff
- Co-training Co-training
- Counterfactual LTR Counterfactual LTR
- Covariance Covariance
- CPU offload CPU offload
- Credit assignment Credit assignment
- Crew Crew
- Critic agent Critic agent
- Cross-attention Cross-attention
- Cross-encoder Cross-encoder
- Cross-entropy math Cross-entropy math
- Cross-modal retrieval Cross-modal retrieval
- Crowdsourcing Crowdsourcing
- CUDA CUDA
- CUDA graph CUDA graph
- cuDNN cuDNN
- Curriculum learning Curriculum learning
- Curse of dimensionality Curse of dimensionality
- CutMix CutMix
- CVR CVR
- Cyber offense risk Cyber offense risk
- SAC SAC
- Safety classifier Safety classifier
- Saliency map Saliency map
- SAM SAM
- SAM optimizer SAM optimizer
- Sampling Sampling
- Sandbox Sandbox
- SARSA SARSA
- Scaffolding Scaffolding
- Scalable oversight Scalable oversight
- ScaNN ScaNN
- Schema drift Schema drift
- Scientific literature mining Scientific literature mining
- Score-based model Score-based model
- Scratchpad Scratchpad
- Seasonality Seasonality
- Secure aggregation Secure aggregation
- Seed Seed
- Self-ask Self-ask
- Self-attention Self-attention
- Self-BLEU Self-BLEU
- Self-consistency Self-consistency
- Self-play Self-play
- Self-RAG Self-RAG
- Self-refine Self-refine
- Self-supervised learning Self-supervised learning
- Semantic cache Semantic cache
- Semantic role labeling Semantic role labeling
- Semantic segmentation Semantic segmentation
- Semi-supervised learning Semi-supervised learning
- Sensitivity Sensitivity
- SentencePiece SentencePiece
- Sentiment analysis Sentiment analysis
- SEP token SEP token
- Seq2Seq Seq2Seq
- Sequential recommendation Sequential recommendation
- Serendipity Serendipity
- Serving Serving
- Session-based recommendation Session-based recommendation
- Session search Session search
- SFT SFT
- SGD SGD
- Shadow deployment Shadow deployment
- SHAP SHAP
- Sharpness-Aware Minimization Sharpness-Aware Minimization
- Short-term memory Short-term memory
- Sigmoid Sigmoid
- SiLU SiLU
- Silver labels Silver labels
- Sim2Real Sim2Real
- Sim2Real gap Sim2Real gap
- SimCLR SimCLR
- SimHash SimHash
- Simpson's paradox Simpson's paradox
- Single-cell ML Single-cell ML
- SIQA SIQA
- Skip connection Skip connection
- SLA SLA
- SLAM SLAM
- Slice-based evaluation Slice-based evaluation
- Sliding window attention Sliding window attention
- Slot filling Slot filling
- Smoke eval Smoke eval
- SMOTE SMOTE
- SMT solver SMT solver
- Snapshot ensembles Snapshot ensembles
- Snorkel Snorkel
- Socioeconomic impact Socioeconomic impact
- Softmax Softmax
- Softmax распределение Softmax распределение
- Soft parameter sharing Soft parameter sharing
- Softplus Softplus
- Soft prompts Soft prompts
- Sound event detection Sound event detection
- SoundStream SoundStream
- Span corruption Span corruption
- Sparse retrieval Sparse retrieval
- Sparse reward Sparse reward
- Speaker diarization Speaker diarization
- Speaker embedding Speaker embedding
- Special tokens Special tokens
- Specification gaming Specification gaming
- Specificity Specificity
- Speculative decoding Speculative decoding
- Speculative execution Speculative execution
- Speech-to-text Speech-to-text
- Spell correction Spell correction
- Split learning Split learning
- SQL generation SQL generation
- SQuAD SQuAD
- Stable Diffusion Stable Diffusion
- State machine State machine
- State Space Model State Space Model
- Static analysis + LLM Static analysis + LLM
- Stationarity Stationarity
- Stationarity test Stationarity test
- Stemming Stemming
- Stereo vision Stereo vision
- Stochastic depth Stochastic depth
- Stochastic gradient descent Stochastic gradient descent
- Stochastic Weight Averaging Stochastic Weight Averaging
- Stop sequences Stop sequences
- Stop words Stop words
- Streaming Streaming
- Stride Stride
- Structured output Structured output
- Structured pruning Structured pruning
- Student model Student model
- Style transfer Style transfer
- Subword Subword
- Summarization Summarization
- Super-resolution Super-resolution
- Supervisor pattern Supervisor pattern
- Survival analysis Survival analysis
- SVD SVD
- SVM SVM
- Swarm Swarm
- SWE-bench SWE-bench
- Swin Transformer Swin Transformer
- Swish Swish
- Sycophancy Sycophancy
- Symbolic execution Symbolic execution
- Synthetic data Synthetic data
- Synthetic data generation Synthetic data generation
- Synthetic labels Synthetic labels
- System prompt System prompt
- Сверточная сеть Сверточная сеть
- Сегментация Сегментация
- Седловая точка Седловая точка
- Скалярное произведение Скалярное произведение
- Скрытый слой Скрытый слой
- Случайный лес Случайный лес
- Состояние Состояние
- Стратификация Стратификация
- Стэкинг Стэкинг
- Сэмплирование Сэмплирование
Т
- T5 T5
- Table extraction Table extraction
- TabNet TabNet
- Tabular deep learning Tabular deep learning
- Tactile sensing Tactile sensing
- Talking head Talking head
- Tanh Tanh
- Target encoding Target encoding
- Target leakage Target leakage
- Target leakage in time Target leakage in time
- Task success rate Task success rate
- Taxonomy Taxonomy
- TD3 TD3
- Teleoperation Teleoperation
- Temporal convolution Temporal convolution
- Temporal ensembling Temporal ensembling
- Temporal Fusion Transformer Temporal Fusion Transformer
- Temporal split Temporal split
- Tensor Core Tensor Core
- Tensor parallel inference Tensor parallel inference
- Tensor parallelism Tensor parallelism
- TensorRT TensorRT
- Terminal agent Terminal agent
- Text-audio alignment Text-audio alignment
- Text-to-speech Text-to-speech
- Text-to-SQL Text-to-SQL
- TextVQA TextVQA
- TF-IDF TF-IDF
- TFLite TFLite
- TGI TGI
- Thompson sampling Thompson sampling
- Throughput Throughput
- Thumbs up/down Thumbs up/down
- Time series forecasting Time series forecasting
- Token billing Token billing
- Token dropping Token dropping
- Tokenizer Tokenizer
- Tokens per joule Tokens per joule
- Tokens per second Tokens per second
- Tool calling loop Tool calling loop
- Toolformer Toolformer
- Toolformer prompting Toolformer prompting
- Tool use Tool use
- Topic modeling Topic modeling
- Top-k sampling Top-k sampling
- Top-p Top-p
- torch.compile torch.compile
- TorchServe TorchServe
- Toxicity Toxicity
- Toxicity rate Toxicity rate
- TPU TPU
- Tracing Tracing
- Training-serving skew Training-serving skew
- Trajectory Trajectory
- Trajectory logging Trajectory logging
- Transfer learning Transfer learning
- Transformer-XL Transformer-XL
- Transparency Transparency
- Transposed convolution Transposed convolution
- TREC DL TREC DL
- Tree of Thoughts Tree of Thoughts
- Trend Trend
- Triplet loss Triplet loss
- Triton Triton
- Tri-training Tri-training
- TriviaQA TriviaQA
- Trojan model Trojan model
- TRPO TRPO
- Trustworthy AI Trustworthy AI
- TruthfulQA TruthfulQA
- t-SNE t-SNE
- TTFT TTFT
- TTS TTS
- Two-stage retrieval Two-stage retrieval
- Two-tower model Two-tower model
- Type inference Type inference
- Typical sampling Typical sampling
- Температура Температура
- Тензор Тензор
- Токен Токен
- Токенизация BPE Токенизация BPE
- Трансформер Трансформер
- Трейн/вал/тест Трейн/вал/тест
У
Ф
- F1-мера F1-мера
- Face detection Face detection
- Face recognition Face recognition
- Fairness Fairness
- Faiss Faiss
- Faithfulness Faithfulness
- Falcon Falcon
- Fallback heuristics Fallback heuristics
- Fallback model Fallback model
- Faster R-CNN Faster R-CNN
- Feature distillation Feature distillation
- Feature engineering Feature engineering
- Feature flag Feature flag
- Feature map Feature map
- Feature pyramid Feature pyramid
- Feature selection Feature selection
- Feature store Feature store
- Feature visualization Feature visualization
- Federated learning Federated learning
- Feedback loop Feedback loop
- FEVER FEVER
- Few-shot Few-shot
- Few-shot exemplar selection Few-shot exemplar selection
- FID FID
- Fill-in-the-middle Fill-in-the-middle
- Filter bubble Filter bubble
- FinQA FinQA
- FixMatch FixMatch
- Flamingo Flamingo
- FlashAttention FlashAttention
- FlashDecoding FlashDecoding
- FlexMatch FlexMatch
- Flickr30k Flickr30k
- Flow Matching Flow Matching
- Fluency score Fluency score
- Focal loss Focal loss
- Force control Force control
- Forced alignment Forced alignment
- Forecast horizon Forecast horizon
- Formal verification Formal verification
- Fourier transform Fourier transform
- FP16 FP16
- FP8 FP8
- FPN FPN
- Frame sampling Frame sampling
- Frequency penalty Frequency penalty
- Frequentist statistics Frequentist statistics
- Frozen weights Frozen weights
- FSDP FSDP
- FT-Transformer FT-Transformer
- Full fine-tuning Full fine-tuning
- Full FT after probe Full FT after probe
- Fully connected layer Fully connected layer
- Function calling Function calling
- Файн-тюнинг Файн-тюнинг
Х
- Hallucination rate Hallucination rate
- Handwriting recognition Handwriting recognition
- Hard negatives Hard negatives
- Hard parameter sharing Hard parameter sharing
- Hard positives Hard positives
- Hate speech detection Hate speech detection
- HBM HBM
- He initialization He initialization
- HellaSwag HellaSwag
- Hierarchical clustering Hierarchical clustering
- HiFi-GAN HiFi-GAN
- Highway network Highway network
- Hinge loss Hinge loss
- Hint loss Hint loss
- Hit Rate Hit Rate
- HNSW HNSW
- Hold-out Hold-out
- Horovod Horovod
- HotpotQA HotpotQA
- Huber loss Huber loss
- HuBERT HuBERT
- Human approval gate Human approval gate
- HumanEval HumanEval
- HumanEval+ HumanEval+
- Human evaluation Human evaluation
- Human-in-the-loop Human-in-the-loop
- Humanoid control Humanoid control
- Human oversight Human oversight
- Human preference winrate Human preference winrate
- Hybrid AI system Hybrid AI system
- Hybrid BM25+vector Hybrid BM25+vector
- Hybrid search Hybrid search
- HyDE HyDE
- Hydra Hydra
- Hypothesis testing Hypothesis testing
- Hypothetical document embeddings Hypothetical document embeddings
Ш
Э
Поделиться