T
139 words and 458 other names, in italics, pointing to their entries.
- T2I → text-to-image model
- T2I-Adapter → ControlNet
- T2V → text-to-video model
- T5
- table of confusion → confusion matrix
- taboo search → tabu search
- TabPFN → tabular foundation model
- tabu list → tabu search
- tabu search
- tabular data
- tabular dataset → tabular data
- tabular deep learning → tabular data
- tabular foundation model
- tabular in-context learning → tabular foundation model
- tabular machine learning → tabular data
- tabular Q-learning → Q-learning
- tacit dimension → Polanyi's paradox
- tacit knowledge → Polanyi's paradox
- Tacotron → text-to-speech
- tactic → interactive theorem proving
- tactile sensing
- tactile sensor → tactile sensing
- tagset → part-of-speech tagging
- TAI → transformative AI
- tail → long tail
- tail latency → latency
- Taiwan Semiconductor Manufacturing Company → TSMC
- TAKE IT DOWN Act
- takeoff
- take-off → takeoff
- takeoff speed → takeoff
- Talkie → MiniMax (company)
- Taming Transformers → VQGAN
- tamper-resistant safeguards → fine-tuning attack
- tanh
- tanh function → tanh
- Tanimoto coefficient → Jaccard similarity
- TAPT → continued pretraining
- target domain → domain adaptation
- target encoding → one-hot encoding
- target label → label
- target leakage → data leakage
- target network
- target policy → off-policy
- target Q-network → target network
- target shift → label shift
- targeted LAT → latent adversarial training
- targeted latent adversarial training → latent adversarial training
- task → task (machine learning)
- task arithmetic
- task automation → automation
- task decomposition
- task environment → environment
- task exposure
- task (machine learning)
- task vector → task arithmetic
- task vectors → task arithmetic
- task-adaptive pretraining → continued pretraining
- task-completion time horizon → METR time horizon
- task-specific AI → narrow AI
- τ → trajectory
- tau2-bench → τ-bench
- τ-bench
- τ²-bench → τ-bench
- Tay
- Tay (chatbot) → Tay
- Taylor expansion → Taylor series
- Taylor polynomial → Taylor series
- Taylor series
- Taylor's theorem → Taylor series
- TayTweets → Tay
- TBD Lab → Meta Superintelligence Labs
- TBench → Terminal-Bench
- TBox → description logic
- TCAV → testing with concept activation vectors
- t-closeness → k-anonymity
- TD error → temporal difference learning
- TD learning → temporal difference learning
- TD(0) → temporal difference learning
- TD3 → deep deterministic policy gradient
- TDA → data attribution
- TD-Gammon
- TDIDT → ID3
- t-distributed stochastic neighbor embedding → t-SNE
- TD(λ) → temporal difference learning
- TDM exception → text and data mining exception
- TDM opt-out → text and data mining exception
- TDNN → time delay neural network
- teacher forcing
- teacher model → knowledge distillation
- teacher-forced training → teacher forcing
- teacher–student training → knowledge distillation
- tech solutionism → techno-solutionism
- technological singularity
- technological solutionism → techno-solutionism
- technological unemployment
- technology node → process node
- techno-solutionism
- technosolutionism → techno-solutionism
- TEE → trusted execution environment
- Max Tegmark
- teleop → teleoperation
- teleoperate → teleoperation
- teleoperated → teleoperation
- teleoperation
- telepresence → teleoperation
- teleprompter → DSPy
- telerobotics → teleoperation
- temp → temperature
- temperature
- tempered overfitting → benign overfitting
- template filling → information extraction
- templated prompt → prompt template
- temporal abstraction → hierarchical reinforcement learning
- temporal action localization → action recognition
- temporal credit assignment → credit assignment problem
- temporal difference learning
- temporal logic
- tense logic → temporal logic
- tensor
- tensor compiler → AI compiler
- tensor core
- tensor cores → tensor core
- tensor model parallelism → tensor parallelism
- tensor parallelism
- Tensor Processing Unit → TPU
- tensor slicing → tensor parallelism
- TensorBoard → TensorFlow
- TensorFlow
- TensorFlow 2 → TensorFlow
- TensorFlow Hub → model hub
- TensorFlow Lite → LiteRT
- TensorFlow Serving → TensorFlow
- TensorFlow.js → TensorFlow
- TensorRT
- TensorRT-LLM
- tensors → tensor
- tera operations per second → TOPS
- teraFLOPS → FLOPS
- term frequency–inverse document frequency → TF-IDF
- term vector model → vector space model
- terminal goal
- terminal phase of training → neural collapse
- terminal value → terminal goal
- Terminal-Bench
- Terminal-Bench 2.0 → Terminal-Bench
- termination condition → stop condition
- terminological logic → description logic
- ternary LLM → BitNet
- Terrence J. Sejnowski → Terry Sejnowski
- Terrence Sejnowski → Terry Sejnowski
- Terry Allen Winograd → Terry Winograd
- Gerald Tesauro
- TESCREAL
- TESCREAL bundle → TESCREAL
- TESCREALism → TESCREAL
- TESCREAList → TESCREAL
- Tesla
- Tesla Autopilot → Tesla
- Tesla, Inc. → Tesla
- Tesler's theorem → AI effect
- test awareness → evaluation awareness
- test data → test set
- test harness → harness
- test set
- test set contamination → data contamination
- test split → test set
- test-clean → LibriSpeech
- testing with concept activation vectors
- test-other → LibriSpeech
- test-time adaptation → test-time training
- test-time attack → evasion attack
- test-time augmentation → data augmentation
- test-time compute
- test-time fine-tuning → test-time training
- test-time scaling → test-time compute
- test-time training
- Texas AI Act → Texas Responsible Artificial Intelligence Governance Act
- Texas HB 149 → Texas Responsible Artificial Intelligence Governance Act
- Texas Responsible Artificial Intelligence Governance Act
- text and data mining exception
- text categorization → text classification
- text classification
- text completion → completion
- text corpus → corpus
- text diffusion → diffusion language model
- text embedding
- text encoder → embedding model
- text generation → natural language generation
- text recognition → optical character recognition
- Text REtrieval Conference → TREC
- Text Retrieval Conference → TREC
- text splitting → chunking
- text summarization
- text tokenizer → tokenizer
- text watermarking → AI watermarking
- text-guided image editing → instruction-based image editing
- text-prompted detection → open-vocabulary detection
- text-to-3D → score distillation sampling
- text-to-3D generation → score distillation sampling
- text-to-code → code generation
- text-to-image generation → text-to-image model
- text-to-image model
- text-to-music → music generation
- text-to-speech
- text-to-SQL → semantic parsing
- text-to-text → T5
- Text-to-Text Transfer Transformer → T5
- text-to-video generation → text-to-video model
- text-to-video model
- textual entailment → natural language inference
- textual inversion
- textual inversion embedding → textual inversion
- TF → TensorFlow
- TF32 → FP32
- TFAIA → SB 53
- TF-IDF
- .tflite → LiteRT
- TFLite → LiteRT
- TFM → tabular foundation model
- TF-TRT → TensorRT
- The AI Scientist → AI scientist
- the algorithm → algorithm
- the alignment problem → alignment problem
- the arXiv → arXiv
- the bitter lesson
- the Blip → OpenAI leadership crisis (2023)
- the ChatGPT moment for robotics → ChatGPT moment
- the GPU-poor → GPU poor
- the hardware lottery → hardware lottery
- the hottest new programming language is English → English is the hottest new programming language
- the Hub → Hugging Face
- the intelligence curse → intelligence curse
- the liar's dividend → liar's dividend
- The New York Times Co. v. Microsoft Corp. → New York Times v. OpenAI
- The Pile
- the Project → Situational Awareness (essay)
- the sharp left turn → sharp left turn
- the singularity → technological singularity
- The Society of Mind
- The Stack
- The Stack v2 → The Stack
- the transformer paper → Attention Is All You Need
- the Turing Trap → Turing trap
- the Turk → Mechanical Turk
- the wall → scaling wall
- Theano
- theorem prover → automated theorem proving
- theoretical receptive field → receptive field
- theory of belief functions → Dempster–Shafer theory
- theory of games → game theory
- theory of mind
- theory-heavy approach → indicator properties of consciousness
- therapy chatbot → AI therapy
- there is no wall → scaling wall
- thin wrapper → wrapper
- things and stuff → panoptic segmentation
- think–act–observe loop → agent loop
- thinking about thinking → metacognition
- thinking between tool calls → interleaved thinking
- thinking blocks → extended thinking
- thinking budget
- Thinking Machines → Thinking Machines Lab
- Thinking Machines Lab
- thinking mode → extended thinking
- thinking model → reasoning model
- thinking tokens → reasoning trace
- thinking trace → reasoning trace
- third-generation neural network → spiking neural network
- third-party AI evaluation
- third-party conformity assessment → notified body
- third-party evaluation → third-party AI evaluation
- this looks like that → prototypical part network
- Thomas J. Watson Research Center → IBM Research
- Thomas Mitchell → Tom Mitchell
- Thompson sampling
- thoughts → reasoning trace
- threat model
- threat modeling → threat model
- threat modelling → threat model
- Three Laws of Robotics
- three Vs → big data
- three-dimensional reconstruction → 3D reconstruction
- three-term controller → PID controller
- threshold logic unit → McCulloch–Pitts neuron
- threshold moving → decision threshold
- threshold tuning → decision threshold
- throttling → rate limit
- throughput
- through-silicon via → advanced packaging
- Sebastian Thrun
- tied embeddings → unembedding
- tied weights → weight sharing
- Tikhonov regularization → ridge regression
- tiktoken → tokenizer
- tile → chiplet
- time between tokens → time per output token
- time complexity → big O notation
- time delay neural network
- time horizon → METR time horizon
- time per output token
- time series → time series forecasting
- time series forecasting
- time series foundation model
- time to first token
- time–frequency representation → spectrogram
- TimeGPT → time series foundation model
- timelines → AI timelines
- time-series cross-validation → cross-validation
- TimesFM → time series foundation model
- time-to-event analysis → survival analysis
- time-to-event model → survival analysis
- Tinker → Thinking Machines Lab
- tiny machine learning → TinyML
- tiny recursive model
- TinyML
- TMLR → JMLR
- TNR → specificity
- token
- token bias → logit bias
- token eviction → KV cache compression
- token generation phase → decode phase
- token log probability → log probability
- token streaming → streaming
- token vocabulary → vocabulary
- token-free model → byte-level model
- tokenisation → tokenization
- tokeniser → tokenizer
- tokenization
- tokenize → tokenization
- tokenizer
- tokenizer-free model → byte-level model
- tokenizing → tokenization
- tokenmaxx → tokenmaxxing
- tokenmaxxer → tokenmaxxing
- tokenmaxxing
- tokens → token
- tokens per minute → rate limit
- tokens per parameter
- tokens per second
- tokens-per-parameter ratio → tokens per parameter
- tokens/s → tokens per second
- token-to-parameter ratio → tokens per parameter
- tok/s → tokens per second
- ToM → theory of mind
- Tom M. Mitchell → Tom Mitchell
- Tongyi → Alibaba
- Giulio Tononi
- tool
- tool AI
- tool AIs → tool AI
- tool call
- tool calling → function calling
- tool center point → end effector
- tool definition → tool
- tool ergonomics → agent–computer interface
- tool invocation → tool call
- tool learning → tool use
- tool loop → agent loop
- tool permissions → permission system
- tool poisoning
- tool poisoning attack → tool poisoning
- tool request → tool call
- tool schema → tool
- tool shadowing → tool poisoning
- tool use
- tool use block → tool call
- tool-augmented language model → augmented language model
- tool-augmented LLM → tool use
- Toolformer
- tools → tool
- Tools to Address Known Exploitation by Immobilizing Technological Deepfakes on Websites and Networks Act → TAKE IT DOWN Act
- tool-using agent → tool use
- top logprobs → log probability
- top-down induction of decision trees → ID3
- topic classification → text classification
- topic model
- topic modeling → topic model
- topic modelling → topic model
- top-k → top-k sampling
- top-k accuracy → accuracy
- top-k routing → router
- TopK SAE → sparse autoencoder
- top-k sampling
- top-k truncation → top-k sampling
- top-p → top-p sampling
- top-p sampling
- TOPS
- TOPS/W → TOPS
- torch → PyTorch
- torch.compile → PyTorch
- TorchScript → PyTorch
- Torch-TensorRT → TensorRT
- Torment Nexus
- Toronto Book Corpus → BookCorpus
- ToT → tree of thoughts
- total parameters → parameter count
- touch sensing → tactile sensing
- toy model
- toy models → toy model
- toy setting → toy model
- TP → tensor parallelism
- TPE → Bayesian optimization
- TPM → rate limit
- TPOT → time per output token
- TPR → recall
- TPS → tokens per second
- TPU
- TPU pod → TPU
- TPU-hour → GPU-hour
- trace a circuit → circuit tracing
- trace decay → eligibility trace
- tracking-by-detection → object tracking
- traffic mirroring → shadow deployment
- TRAIGA → Texas Responsible Artificial Intelligence Governance Act
- train → training
- train data → training data
- train loop → training loop
- train set → training set
- trainable parameter → parameter
- trained → training
- trained to convergence → convergence
- training
- training cluster → AI supercomputer
- training computation → training compute
- training compute
- training compute threshold → compute threshold
- training corpus → corpus
- training cutoff → knowledge cutoff
- training data
- training data attribution → data attribution
- training data cutoff → knowledge cutoff
- training data extraction
- training data extraction attack → training data extraction
- training data licensing → data licensing
- training data poisoning → data poisoning
- training data set → training data
- training dataset → training set
- training environment → RL environment
- training epoch → epoch
- training example → example
- training FLOP → training compute
- training FLOPs → training compute
- training game → training gaming
- training gaming
- training goodput → goodput
- training instability
- training loop
- training loss curve → loss curve
- training opt-out → opt-out
- training pipeline → machine learning pipeline
- training run
- training sample → training set
- training set
- training split → training set
- training step
- training–serving skew
- train-serve skew → training–serving skew
- train/serve skew → training–serving skew
- train–test gap → generalization gap
- train–test leakage → data leakage
- train–test overlap → data contamination
- trajectory
- trajectory generation → trajectory optimization
- trajectory optimisation → trajectory optimization
- trajectory optimization
- trajectory transformer → decision transformer
- Tralalero Tralala → Italian brainrot
- Transactions on Machine Learning Research → JMLR
- transcoder
- transcription → speech recognition
- TransE → knowledge graph embedding
- transfer attack → black-box attack
- transfer function → activation function
- transfer learning
- transformative AI
- transformative artificial intelligence → transformative AI
- transformer
- transformer architecture → transformer
- transformer block
- Transformer Circuits → Transformer Circuits Thread
- Transformer Circuits Thread
- transformer layer → transformer block
- transformer model → transformer
- transformer-circuits.pub → Transformer Circuits Thread
- TransformerLens
- transformer_lens → TransformerLens
- Transformers
- transformers library → Transformers
- Transformer-XL
- transhumanism
- transhumanist → transhumanism
- transition matrix → Markov chain
- translation invariance → invariance
- Translatotron → speech-to-speech model
- transparency (AI) → AI transparency
- Transparency in Frontier Artificial Intelligence Act → SB 53
- transparency obligations
- transposed convolution
- trapezoidal schedule → warmup-stable-decay schedule
- traveling salesman problem
- traveling salesperson problem → traveling salesman problem
- travelling salesman problem → traveling salesman problem
- treacherous turn
- TREC
- TREC track → TREC
- tree all-reduce → all-reduce
- tree of thoughts
- tree pruning → pruning
- tree search → state space search
- tree-based genetic programming → genetic programming
- tree-based model → decision tree
- tree-of-thought prompting → tree of thoughts
- TreeSHAP → SHAP
- tree-structured Parzen estimator → Bayesian optimization
- trial → episode
- triangular mask → causal mask
- trigger word → wake word
- trigram model → n-gram model
- trillion operations per second → TOPS
- triple → knowledge graph
- triple store → Resource Description Framework
- triplet loss → metric learning
- Triton
- Triton kernel → Triton
- Triton language → Triton
- TRM → tiny recursive model
- trojan → backdoor attack
- trojan attack → backdoor attack
- trolley dilemma → trolley problem
- trolley problem
- trolleyology → trolley problem
- trough of disillusionment → hype cycle
- TRPO → trust region policy optimization
- true lift modeling → uplift modeling
- true negative rate → specificity
- true positive rate → recall
- true positive rate parity → equal opportunity
- true risk → empirical risk minimization
- truncatable embedding → Matryoshka embedding
- truncated backpropagation through time → backpropagation through time
- truncated BPTT → backpropagation through time
- truncated SVD → low-rank approximation
- trust and safety → content moderation
- trust region policy optimization
- trusted edit → trusted editing
- trusted editing
- trusted execution environment
- trusted monitor → trusted monitoring
- trusted monitoring
- trustworthy AI
- trustworthy artificial intelligence → trustworthy AI
- truth maintenance system → non-monotonic reasoning
- truthful → truthfulness
- truthful AI → truthfulness
- Truthful QA → TruthfulQA
- truthfulness
- TruthfulQA
- TSFM → time series foundation model
- TSMC
- t-SNE
- tSNE → t-SNE
- TSP → traveling salesman problem
- TSV → advanced packaging
- TTC → test-time compute
- TTFT → time to first token
- TTS → text-to-speech
- TTT → test-time training
- TTT layer → test-time training
- tubelet embedding → patch embedding
- tuned lens
- Tung Tung Tung Sahur → Italian brainrot
- tuning → hyperparameter tuning
- tuning parameter → hyperparameter
- tuning set → validation set
- turbo model → diffusion distillation
- Alan Turing
- Turing Award
- Turing completeness → Turing machine
- Turing machine
- Turing test
- Turing trap
- Turing-complete → Turing machine
- Turing's test → Turing test
- turn detection → voice activity detection
- Turtle → Resource Description Framework
- twin delayed DDPG → deep deterministic policy gradient
- twin network → Siamese network
- twin neural network → Siamese network
- two-armed bandit → multi-armed bandit
- two-class classification → binary classification
- two-part code → minimum description length
- two-stage detector → object detection
- two-tower architecture → two-tower model
- two-tower model
- two-tower retrieval → two-tower model
- typed output → structured outputs