Alignment (Alignment)
Alignment
Kode ti sara si akode ti AI, akode ti output, na akode ti comportement ti AI system ayeke na ndo ti akode ti zo na akode ti valeur. So ayeke kota mingi na yâ ti akode ti système so alingbi ti sara akode so ayeke na yâ ti akode ti zo.
Example: Ti sara si mbeti ti chatbot teti mental health support ayeke sara akode ti sioni na ndo ti akode ti prompt.
Application Programming Interface (API) (Application Programming Interface (API))
Application Programming Interface (API)
Akode ti règle na protocole so azi na akode ti logiciel ti communiquer na ti changer data.
Example: Ti sara API ti OpenAI ti zia prompt na ti wara réponse ti language model na yâ ti web app ti mo.
Artificial General Intelligence (AGI) (Artificial General Intelligence (AGI))
Artificial General Intelligence (AGI)
Forme ti AI so alingbi ti sara akode ti intellectuel so zo alingbi ti sara. A yeke sara akode ti apprendre na ndo ti akode ti domaine.
Example: AGI system alingbi ti apprendre composition ti music, ti sara chirurgie, na ti wara examen ti philosophie sans programmation ti task spécifique.
Artificial Intelligence (AI) (Artificial Intelligence (AI))
Artificial Intelligence (AI)
Simulation ti intelligence ti zo na yâ ti amachine so a programme ti pensé, ti raisonné, na ti sara akode ti autonome.
Example: AI ayeke sara akode ti assistant personnel tongana Siri na akode ti système ti auto-conduite tongana Tesla Autopilot.
AI Ethics (AI Ethics)
AI Ethics
Discipline so ayeke na ndo ti akode ti moral ti développement na ti use ti AI, so ayeke na yâ ti akode ti fairness, privacy, accountability, na non-discrimination.
Example: Ti sara akode ti guideline ti kanga hiring algorithms ti sara discrimination na ndo ti gender wala ethnicity.
Augmented Intelligence (Augmented Intelligence)
Augmented Intelligence
Modèle ti collaboration so AI ayeke sara akode ti améliorer intelligence ti zo na ndo ti ti changer ni.
Example: AI-powered radiology tools so ayeke sara akode ti highlight anomalies teti adocteur, so ayeke sara final diagnosis.
Autonomous Agent (Autonomous Agent)
Autonomous Agent
AI system so alingbi ti sara akode ti desizion ti ni mveni na ti sara akode ti wara akode ti ni sans intervention ti zo.
Example: Robot ti auto-conduite so ayeke sara akode ti naviguer na yâ ti akode ti ville na ti kanga akode ti obstacle na ni mveni.
Backpropagation (Backpropagation)
Backpropagation
Technique ti training neural networks na ti changer weights na reverse ti output na input layers, ti minimiser prediction errors.
Example: A yeke sara ni na yâ ti training image classifiers ti kanga error rate na yâ ti akode ti reconnaître handwritten digits.
Bias (Algorithmic Bias) (Bias (Algorithmic Bias))
Bias (Algorithmic Bias)
Favouritism ti AI outcomes so ayeke na yâ ti akode ti imbalanced wala non-representative training data.
Example: Facial recognition system so ayeke sara akode ti misidentify azo ti couleur mingi mingi na ndo ti underrepresentation na yâ ti training data.
Big Data (Big Data)
Big Data
Akode ti dataset so ayeke kota mingi so ayeke na besoin ti akode ti outil ti store, ti analyser, na ti wara valeur, so a yeke sara ni mingi ti training AI models.
Example: Ti sara akode ti million ti user interactions ti training recommendation engines teti e-commerce platforms.
Black Box Model (Black Box Model)
Black Box Model
Type ti AI wala machine learning model so internal logic ti ni ayeke ngangu ti comprendre na azo, so ayeke sara si a yeke ngangu ti comprendre akode ti desizion.
Example: Deep neural network so a yeke sara ni ti approve loans me so ayeke sara akode ti explication ti polele teti so mbeni applicant a wara ni na mbeni a kanga ni.
Cognitive Computing (Cognitive Computing)
Cognitive Computing
AI systems so a sara ni ti simuler akode ti pensée ti zo, tongana reasoning na learning, na ti sara akode ti technique tongana NLP na pattern recognition.
Example: Cognitive computing system so azi na akode ti professionnel ti legal ti analyser case law na ti prédire outcomes.
Computer Vision (Computer Vision)
Computer Vision
Domaine ti artificial intelligence so azi na akode ti ordinateur ti interpréter na ti process visual data tongana images na video.
Example: Facial recognition systems so ayeke sara akode ti identifier azo na yâ ti security footage na ti sara computer vision.
Corpus (Corpus)
Corpus
Kota collection ti written wala spoken texts so a yeke sara ni ti training language models.
Example: Common Crawl dataset ayeke public web corpus so a yeke sara ni ti training large language models tongana GPT.
Data Drift (Data Drift)
Data Drift
Phénomène so input data ayeke changer na ndo ti temps, so ayeke sara si model performance ayeke kanga.
Example: Predictive maintenance model teti industrial equipment ayeke kanga accuracy ti ni na ti sara si new sensor technology a zia ni.
Data Labelling (Data Labelling)
Data Labelling
Kode ti annoter data na tags wala labels ti sara si a yeke nzoni teti supervised learning.
Example: Labelling thousands of tumour images tongana benign wala malignant ti training cancer detection model.
Data Mining (Data Mining)
Data Mining
Kode ti découvrir meaningful patterns, correlations, na anomalies na yâ ti large datasets.
Example: Retailers so ayeke sara data mining ti identifier so azo so ayeke sara nappies ayeke sara bière.
Deep Learning (Deep Learning)
Deep Learning
Subfield ti machine learning so ayeke sara multi-layered neural networks ti model complex patterns na yâ ti data.
Example: Deep learning a yeke sara ni na yâ ti language models tongana GPT-4 na image generation models tongana Stable Diffusion.
Diffusion Models (Diffusion Models)
Diffusion Models
Classe ti generative models so ayeke apprendre ti sara data na ti changer random noise na structured outputs.
Example: Stable Diffusion ayeke sara photorealistic images na text prompts na ti sara diffusion techniques.
Embedding (Embedding)
Embedding
Numerical vector representation ti data, so a yeke sara ni mingi ti capture semantic meaning ti words, images, wala sentences.
Example: Na yâ ti NLP, word 'bank' alingbi ti wara similar embeddings na 'money' me different na 'riverbank' na ndo ti context.
Epoch (Epoch)
Epoch
Full iteration na ndo ti entire training dataset na yâ ti training process ti machine learning model.
Example: Si dataset ayeke na 1,000 examples na model ayeke wara ni fani oko na yâ ti training, so ayeke oko epoch.
Ethical AI (Ethical AI)
Ethical AI
Design na deployment philosophy so ayeke sara si AI technologies ayeke sara akode ti transparently, equitably, na na ndo ti societal values.
Example: AI hiring tool so ayeke na yâ ti bias checks ti kanga discrimination na ndo ti minority candidates.
Expert System (Expert System)
Expert System
AI system so ayeke sara akode ti decision-making abilities ti human expert na yâ ti specific domain na ti sara rules na logic.
Example: Expert system so a yeke sara ni na yâ ti agriculture ti recommander crop treatments na ndo ti soil data na pest history.
Explainable AI (XAI) (Explainable AI (XAI))
Explainable AI (XAI)
AI systems so a sara ni ti sara si internal processes na decisions ti ni ayeke comprendre na azo, ti augmenter trust na accountability.
Example: Medical diagnostic AI so ayeke sara akode ti recommendation me so ayeke sara akode ti expliquer so akode ti symptom a sara si a wara conclusion so.
Few-shot Learning (Few-shot Learning)
Few-shot Learning
Machine learning method so model ayeke training wala fine-tuned na ti sara akode ti small number ti labelled examples.
Example: Customising LLM ti sara legal emails na ti sara akode ti 10 examples.
Fine-tuning (Fine-tuning)
Fine-tuning
Kode ti wara pre-trained model na ti sara training ni na new, smaller dataset ti specialise ni teti specific task.
Example: Fine-tuning general LLM tongana GPT na internal legal documents ti sara legal drafting assistant.
Foundation Model (Foundation Model)
Foundation Model
Kota model so a training ni na diverse na broad data so alingbi ti sara akode ti many downstream tasks.
Example: GPT-4 na PaLM 2 ayeke foundation models so alingbi ti sara summarisation, Q&A, translation, na mingi.
Fuzzy Logic (Fuzzy Logic)
Fuzzy Logic
Forme ti logic so ayeke sara akode ti approximate values na ndo ti fixed true/false (binary) logic, so ayeke nzoni teti reasoning na ndo ti uncertainty.
Example: A yeke sara ni na yâ ti climate control systems ti adjust temperature na ndo ti fuzzy inputs tongana 'a bit hot' wala 'very cold'.
Generative Adversarial Network (GAN) (Generative Adversarial Network (GAN))
Generative Adversarial Network (GAN)
Generative model architecture so two networks — generator na discriminator — ayeke sara akode ti améliorer output quality.
Example: GANs a yeke sara ni ti sara deepfake videos wala ti sara realistic product photos na sketches.
Generative AI (Generative AI)
Generative AI
Catégorie ti artificial intelligence so alingbi ti sara new content — tongana text, images, music, wala video — na training data.
Example: ChatGPT so ayeke sara blog posts wala Midjourney so ayeke sara digital artwork na textual prompts.
Generative Pre-trained Transformer (GPT) (Generative Pre-trained Transformer (GPT))
Generative Pre-trained Transformer (GPT)
Classe ti large language models so OpenAI a sara ni so ayeke sara transformer architecture na a pre-trained ni na vast amounts ti text data ti sara variety ti language tasks.
Example: GPT-4 alingbi ti sara essays, ti traduire languages, na ti summariser documents na minimal prompting.
Genetic Algorithm (Genetic Algorithm)
Genetic Algorithm
Optimisation technique so a inspirer ni na natural selection so solutions ayeke évoluer na ndo ti temps na mutation, crossover, na selection.
Example: A yeke sara ni ti design efficient neural network architectures na ti simuler survival of the fittest.
Hallucination (Hallucination)
Hallucination
Génération ti plausible-sounding me factually incorrect wala nonsensical content na AI model.
Example: Language model so ayeke sara non-existent citation wala ti sara false historical facts.
Heuristic (Heuristic)
Heuristic
Practical approach ti problem-solving so ayeke sara si a yeke wara perfect solution me so ayeke nzoni teti immediate goals.
Example: Ti sara rule of thumb ti estimer delivery time na yâ ti logistics AI system.
Hyperparameter (Hyperparameter)
Hyperparameter
Configuration value so a zia ni kozo ti training machine learning model, tongana learning rate wala number ti layers.
Example: Adjusting batch size na 32 na 128 ti améliorer training speed na model performance.
Inference (Inference)
Inference
Kode ti sara trained machine learning model ti sara predictions wala ti sara outputs na new input data.
Example: Ti sara fine-tuned GPT model ti sara emails teti customer support team.
Intent Detection (Intent Detection)
Intent Detection
Task na yâ ti natural language understanding so system ayeke identifier user’s goal wala purpose na yâ ti message.
Example: Na yâ ti chatbot, reconnaître 'I want to book a flight' tongana travel booking intent.
Internet of Things (IoT) (Internet of Things (IoT))
Internet of Things (IoT)
Réseau ti interconnected physical devices so a zia ni na sensors, software, na other technologies ti collect na ti changer data.
Example: Smart thermostats na fridges so ayeke sara usage data na ti adjust settings na ti sara AI analytics.
Interpretability (Interpretability)
Interpretability
Extent so human alingbi ti comprendre internal mechanics ti machine learning model na decision-making process ti ni.
Example: Decision tree ayeke plus interpretable na deep neural network teti so decisions ti ni ayeke traceable.
Jupyter Notebook (Jupyter Notebook)
Jupyter Notebook
Open-source interactive computing environment so azi na users ti sara code, ti visualiser outputs, na ti document analysis na yâ ti single interface.
Example: Data scientists ayeke sara Jupyter Notebooks ti prototype machine learning models na ti share results.
K-Nearest Neighbours (KNN) (K-Nearest Neighbours (KNN))
K-Nearest Neighbours (KNN)
Simple, non-parametric machine learning algorithm so a yeke sara ni teti classification na regression. A yeke sara decisions na ndo ti closest training examples na yâ ti feature space.
Example: Ti classifier new fruit tongana apple wala pear, KNN ayeke check so labelled fruits ayeke closest na yâ ti shape na colour.
Knowledge Graph (Knowledge Graph)
Knowledge Graph
Data structure so ayeke sara nodes na edges ti représenter na ti store interlinked descriptions ti entities na relationships ti ni.
Example: Google's knowledge panel ayeke powered na knowledge graph so ayeke connecter entities tongana people, places, na events.
Language Learning Model Optimisation (LLMO) (Language Learning Model Optimisation (LLMO))
Language Learning Model Optimisation (LLMO)
Techniques so a yeke sara ni ti améliorer performance, efficiency, wala adaptability ti large language models teti specific tasks wala domains.
Example: Ti sara quantisation na instruction tuning ti optimiser LLM teti enterprise use.
Large Language Model (LLM) (Large Language Model (LLM))
Large Language Model (LLM)
Type ti deep learning model so a training ni na vast amounts ti textual data so alingbi ti sara, ti comprendre, na ti raisonner na human language.
Example: ChatGPT na Claude ayeke LLMs so a training ni ti azi na yâ ti writing, coding, na answering questions.
Latent Space (Latent Space)
Latent Space
High-dimensional abstract representation so similar inputs ayeke grouped close together, so a yeke sara ni na yâ ti generative models na embeddings.
Example: Na yâ ti image generation, manipulating latent space alingbi ti changer features tongana brightness wala emotion.
Learning Rate (Learning Rate)
Learning Rate
Key hyperparameter na yâ ti training so ayeke control so model weights ayeke adjusted na ndo ti loss gradient.
Example: High learning rate alingbi ti sara si a yeke overshooting minima, na low rate ayeke sara si training progress ayeke kanga.
Machine Learning (ML) (Machine Learning (ML))
Machine Learning (ML)
Branche ti AI so azi na systems ti apprendre na data na ti améliorer performance sans ti sara explicit programming.
Example: Spam filters ayeke sara machine learning ti classifier emails tongana spam wala non na ndo ti past examples.
Model Drift (Model Drift)
Model Drift
Phénomène so accuracy ti model ayeke kanga na ndo ti temps na ndo ti changes na data wala environment.
Example: Fraud detection model ayeke kanga accuracy ti ni na ti sara si fraud tactics ayeke évoluer.
Model Training (Model Training)
Model Training
Kode ti zia data na machine learning model na ti adjust parameters ti ni ti minimiser error.
Example: Training recommendation engine na customer purchase history ti suggérer new products.
Multimodal AI (Multimodal AI)
Multimodal AI
AI systems so alingbi ti process na ti intégrer multiple types ti data tongana text, images, audio, na video.
Example: Model tongana GPT-4 Vision so alingbi ti lire text na ti interpréter images na même temps.
Natural Language Processing (NLP) (Natural Language Processing (NLP))
Natural Language Processing (NLP)
Subfield ti AI so ayeke na ndo ti interaction entre computers na human (natural) languages. A azi na machines ti lire, ti comprendre, na ti répondre na human language.
Example: NLP a yeke sara ni na yâ ti voice assistants, language translation apps, na chatbots.
Neural Network (Neural Network)
Neural Network
Machine learning model so a inspirer ni na structure ti human brain, so a composer ni na layers ti interconnected nodes (neurons).
Example: Neural networks ayeke na ndo ti deep learning models so a yeke sara ni na yâ ti image na speech recognition.
Noise (Noise)
Noise
Random wala irrelevant information na yâ ti data so alingbi ti kanga meaningful patterns na ti sara sioni na model performance.
Example: Sensor errors wala typo-filled data entries alingbi ti wara ni tongana noise.
Ontology (Ontology)
Ontology
Structured framework so ayeke categoriser na ti définir relationships entre concepts na yâ ti domain, so a yeke sara ni mingi na yâ ti semantic AI systems.
Example: Ontology na yâ ti healthcare alingbi ti définir so symptoms ayeke na ndo ti diseases na treatments.
Overfitting (Overfitting)
Overfitting
Modelling error so machine learning model ayeke capture noise na yâ ti training data na ti sara sioni na new data.
Example: Model so ayeke memoriser training answers me so alingbi ti sara unseen test data ayeke overfitted.
Predictive Analytics (Predictive Analytics)
Predictive Analytics
Kode ti sara data, algorithms, na AI ti identifier likelihood ti future outcomes na ndo ti historical data.
Example: Retailers ayeke sara predictive analytics ti forecast demand teti certain products.
Pre-training (Pre-training)
Pre-training
Kode ti sara initial training ti model na large, general dataset kozo ti fine-tuning ni teti specific tasks.
Example: GPT models ayeke pre-trained na large corpora kozo ti customising ni teti customer service chatbots.
Prompt Engineering (Prompt Engineering)
Prompt Engineering
Art na science ti crafting effective prompts ti steer output ti large language models.
Example: Ti zia system instructions tongana 'Reply as a polite tutor' ayeke example ti prompt engineering.
Quantisation (Quantisation)
Quantisation
Model compression technique so ayeke kanga number ti bits so a yeke sara ni ti représenter weights na activations, ti améliorer efficiency.
Example: Quantising model na 32-bit na 8-bit ayeke améliorer performance na mobile devices.
Quantum Computing (Quantum Computing)
Quantum Computing
New paradigm ti computing so ayeke na ndo ti quantum mechanics, so ayeke na potential teti exponential processing capabilities.
Example: Quantum computing alingbi ti accélérer AI training na ndo ti classical limits.
Reasoning Engine (Reasoning Engine)
Reasoning Engine
System na yâ ti AI so ayeke sara logical conclusions na set ti facts wala data na ti sara rules wala inference algorithms.
Example: AI diagnosis tool ayeke sara reasoning engine ti déduire possible medical conditions na ndo ti symptoms.
Reinforcement Learning (RL) (Reinforcement Learning (RL))
Reinforcement Learning (RL)
Domaine ti machine learning so agents ayeke apprendre na ti sara interaction na environment ti ni ti maximiser cumulative rewards.
Example: Robot so ayeke apprendre ti tambula na trial na error na ti sara RL techniques.
Reinforcement Learning with Human Feedback (RLHF) (Reinforcement Learning with Human Feedback (RLHF))
Reinforcement Learning with Human Feedback (RLHF)
Learning method so human preferences ayeke guide AI's reward signal, so a yeke sara ni mingi na yâ ti fine-tuning language models.
Example: ChatGPT a training ni na RLHF ti sara plus helpful na safe responses.
Retrieval-Augmented Generation (RAG) (Retrieval-Augmented Generation (RAG))
Retrieval-Augmented Generation (RAG)
Method so ayeke combiner information retrieval na generation, so LLM ayeke fetch relevant documents ti améliorer response ti ni.
Example: AI assistant so ayeke retrieve na cite product specs na ti sara generation ti answer na technical question.
Self-Supervised Learning (Self-Supervised Learning)
Self-Supervised Learning
Training approach so model ayeke apprendre patterns na ti sara own labels na raw data, ti kanga reliance na human-annotated data.
Example: BERT a training ni na self-supervised learning na ti prédire missing words na text.
Semantic Search (Semantic Search)
Semantic Search
Search technique so ayeke comprendre user intent na contextual meaning, na ndo ti keyword matching.
Example: Searching 'how to fix a leaking tap' ayeke sara guides même si term 'leaking tap' ayeke na yâ ti document.
Sentiment Analysis (Sentiment Analysis)
Sentiment Analysis
Kode ti identifier emotions, opinions, wala attitudes na text, so a yeke classifier ni mingi tongana positive, negative, wala neutral.
Example: Analysing tweets ti gauge public reaction na new product.
Stochastic (Stochastic)
Stochastic
Involving randomness wala probabilistic behaviour, so a yeke sara ni mingi na yâ ti generative AI na optimisation algorithms.
Example: Output ti GPT-4 ayeke varier teti même input na ndo ti stochastic decoding process ti ni.
Strong AI (Strong AI)
Strong AI
A yeke wara ni tongana Artificial General Intelligence (AGI), so ayeke na ndo ti machines na human-level cognitive capabilities na ndo ti akode ti domains.
Example: Future AI so alingbi ti sara novels, ti planifier cities, na ti résoudre ethical dilemmas equally well.
Super Artificial Intelligence (SAI) (Super Artificial Intelligence (SAI))
Super Artificial Intelligence (SAI)
Theoretical AI so ayeke na ndo ti human intelligence na ndo ti akode ti aspects—reasoning, creativity, emotional intelligence, etc.
Example: SAI alingbi ti sara new sciences na philosophies independently.
Supervised Learning (Supervised Learning)
Supervised Learning
Machine learning technique so models ayeke training na labelled data ti apprendre input-output mappings.
Example: Teaching model ti classifier emails tongana spam wala non na ti sara historical examples.
Synthetic Data (Synthetic Data)
Synthetic Data
Artificially generated data so ayeke simuler real-world data, so a yeke sara ni mingi teti training si real data ayeke scarce wala sensitive.
Example: Creating synthetic medical images ti training diagnostic models sans ti sara violation ti patient privacy.
Token (Token)
Token
Unit ti text so LLMs ayeke process ni—tongana word wala word piece.
Example: Sentence 'Hello world!' a split ni na 3 tokens: 'Hello', 'world', na '!'.
Tokenisation (Tokenisation)
Tokenisation
Kode ti kanga text na tokens teti processing na model.
Example: Na yâ ti NLP, 'ChatGPT is great' ayeke wara ['Chat', 'G', 'PT', 'is', 'great'].
Transfer Learning (Transfer Learning)
Transfer Learning
Ti sara knowledge na oko task ti améliorer learning na mbeni related task, ti kanga training time na data needs.
Example: Fine-tuning model so a training ni na English text ti sara sentiment analysis na mbeni language.
Transformer (Transformer)
Transformer
Neural network architecture so ayeke sara attention mechanisms ti model sequential data, so a yeke sara ni mingi na yâ ti LLMs.
Example: BERT, GPT, na T5 ayeke transformer-based models.
Underfitting (Underfitting)
Underfitting
Si model ayeke simple mingi ti capture patterns na yâ ti training data, so ayeke sara sioni na performance.
Example: Linear model so ayeke sara ti prédire complex image classifications alingbi ti underfit.
Unsupervised Learning (Unsupervised Learning)
Unsupervised Learning
Learning approach so models ayeke identifier patterns wala clusters na unlabelled data.
Example: Grouping customers na ndo ti purchasing behaviour sans predefined labels.
User Intent (User Intent)
User Intent
Goal wala purpose na ndo ti user's query wala interaction.
Example: User so ayeke sara 'how to bake a cake' alingbi ti wara recipe.
Validation Set (Validation Set)
Validation Set
Subset ti data so a yeke sara ni ti évaluer model performance na yâ ti training na ti tune hyperparameters.
Example: A yeke sara ni ti detect overfitting kozo ti final testing.
Vector Database (Vector Database)
Vector Database
Database so a design ni ti store na ti search vector embeddings so a yeke sara ni na yâ ti AI tasks tongana similarity search na RAG.
Example: Pinecone na Weaviate ayeke vector databases teti storing text wala image embeddings.
Vector Embedding (Vector Embedding)
Vector Embedding
Numeric representation ti data so ayeke préserver semantic meaning na relationships na yâ ti vector space.
Example: Words 'king' na 'queen' ayeke na similar embeddings na subtle gender differences.
Virtual Assistant (Virtual Assistant)
Virtual Assistant
AI-powered software agent so azi na users ti sara tasks na conversation wala voice commands.
Example: Siri, Alexa, na Google Assistant ayeke popular virtual assistants.
Voice Recognition (Voice Recognition)
Voice Recognition
Technology so ayeke interpréter na ti changer spoken language na text wala action.
Example: Voice typing na voice commands ayeke na ndo ti voice recognition systems.
Weak AI (Weak AI)
Weak AI
AI systems so a design ni ti sara narrow, specific task sans general intelligence.
Example: Chess-playing AI so alingbi ti comprendre language wala ti conduire car ayeke example ti weak AI.
Web Scraping (Web Scraping)
Web Scraping
Automated extraction ti information na websites, so a yeke sara ni mingi ti collect training data wala ti monitor content.
Example: Scraping real estate listings ti training property valuation model.
Weight (Weight)
Weight
Parameter na yâ ti neural networks so ayeke déterminer strength ti influence oko node ayeke na ndo ti mbeni.
Example: Weights ayeke adjust na yâ ti training ti minimiser model's error.
Whisper (Whisper)
Whisper
Speech-to-text model so OpenAI a sara ni so alingbi ti transcrire audio na multiple languages.
Example: Whisper alingbi ti transcrire lectures na podcasts na high accuracy.
YAML (YAML)
YAML
Human-readable format teti data serialisation, so a yeke sara ni mingi teti configuration files na yâ ti machine learning workflows.
Example: Defining model parameters na YAML file teti training na PyTorch.
Zero-shot Learning (Zero-shot Learning)
Zero-shot Learning
Ability ti model ti sara tasks so a training ni na ndo ti leveraging general knowledge.
Example: Model so ayeke sara legal questions même si a training ni na ndo ti legal data.
Zettabyte (Zettabyte)
Zettabyte
Unit ti digital data so ayeke oko sextillion (10^21) bytes, so a yeke sara ni mingi ti décrire scale ti internet data.
Example: Global internet traffic a wara 1 zettabyte per year na 2016.
