EXPLORE / 06
Terms index
Specialized AI and software-development terms with concise explanations
Entity records
4517 records · 60/113
- 2361
Model Derivatives
New versions of models created by fine-tuning or merging original base models.
Website pending - 2362
Training Recipe
Complete reproducible training plan including data, code, and hyperparameters
Website pending - 2363
Judge Model
Auxiliary models used to evaluate or score other models' outputs
Website pending - 2364
Lossy Self-Improvement
A non-perfect self-upgrading mode characterized by information loss or performance fluctuations during iterative optimization.
Website pending - 2365
Neural Computer
A novel hardware system combining neural networks with traditional computing architectures.
Website pending - 2366
Muon Optimizer
A deep learning optimization algorithm based on orthogonalized momentum for efficiency.
Website pending - 2367
Positive Alignment
An alignment paradigm where AI systems actively pursue human well-being rather than merely avoiding harm.
Website pending - 2368
Technological Singularity
The point where AI surpasses human intelligence, causing unpredictable societal change
Website pending - 2369
Text-based Persuasion
The ability to change audience attitudes or behaviors through written language.
Website pending - 2370
Capability Gap
The difference in performance between models on specific benchmark tasks.
Website pending - 2371
Bitter Lesson
The observation that general methods leveraging computation ultimately outperform hand-coded heuristics.
Website pending - 2372
Decentralized Architecture
A system structure with no single point of control, relying on distributed network collaboration.
Website pending - 2373
Open-weight LLM
Large language models with publicly available weights, allowing local deployment and modification.
Website pending - 2374
Low-regret policy
Policy measures that remain beneficial or harmless regardless of future technological outcomes, minimizing potential downside risks.
Website pending - 2375
Sandbox Isolation
Security technology restricting program execution to a closed environment to prevent impact on external systems
Website pending - 2376
Misaligned Goals
The actual objectives pursued by an AI system are inconsistent with designer intent or human values
Website pending - 2377
Emergent Communication
Communication methods or language structures spontaneously formed by multiple AI agents without pre-defined protocols
Website pending - 2378
Phase Change
Sudden qualitative transformation in a system as parameters cross a critical threshold.
Website pending - 2379
Wall Clock Time
The actual physical time elapsed from the start to the end of a task.
Website pending - 2380
Inference Scaling
Methods to improve model output quality or speed by increasing computational resources during inference.
Website pending - 2381
Synthetic Morphology
Using algorithms and robotics to build novel life forms with self-organizing capabilities.
Website pending - 2382
Physicalism
A philosophical stance holding that all phenomena are ultimately explainable by physical facts.
Website pending - 2383
Multi-agent Swarms
A distributed system architecture where multiple autonomous agents collaborate to complete complex tasks.
Website pending - 2384
Platt Scaling
A post-processing method to convert SVM outputs into probability estimates
Website pending - 2385
Sequential Minimal Optimization
An efficient optimization algorithm for training Support Vector Machines
Website pending - 2386
Agentic Shopping
An automated process where AI agents autonomously handle searching, comparing, and purchasing decisions.
Website pending - 2387
Genomic Language Model (GLM)
Large models based on Transformer architecture designed to process DNA/RNA sequences.
Website pending - 2388
Autoregressive Diffusion
A hybrid architecture combining autoregressive temporal generation with diffusion denoising processes.
Website pending - 2389
Token Share
The proportion of total tokens consumed by a specific model on a platform, reflecting its usage popularity.
Website pending - 2390
Sparse Reconstruction
Technique to recover complete 3D scenes or object geometries from limited observational data.
Website pending - 2391
Long Decode Continuation
An inference optimization technique allowing models to generate extremely long text outputs.
Website pending - 2392
Recursive Language Model
Model architecture handling complex logic via self-invocation or hierarchical structures
Website pending - 2393
Agent Arena
Evaluation platform comparing AI agents' task execution capabilities and costs.
Website pending - 2394
Proximal Point Algorithm
An iterative optimization method solving regularized subproblems to approximate the original problem.
Website pending - 2395
Client Drift
The phenomenon in federated learning where local updates deviate from the global optimum.
Website pending - 2396
Variational Inequality
A generalized optimization framework often used to describe game equilibria or constrained problems.
Website pending - 2397
Activation Steering
Technique controlling model output by modifying internal activation values
Website pending - 2398
Symbolic Equivalence
A method to determine if two mathematical expressions are logically identical
Website pending - 2399
Serialization
The process of converting data structures into linear formats for transmission or storage
Website pending - 2400
Incremental Delivery
A method of splitting large changes into small batches for continuous integration to reduce risk.
Website pending