14 terms
Showing all terms starting with C
A prompting technique that encourages LLMs to reason step-by-step before giving a final answer, improving accuracy on complex tasks.
A software application that simulates conversation with users, powered by rule-based logic or AI language models.
A supervised learning task where an AI assigns input data to one of several predefined categories or labels.
Contrastive Language-Image Pretraining - an OpenAI model that learns visual concepts from natural language descriptions, enabling text-to-image search.
A field of AI enabling machines to interpret and make decisions based on visual data such as images and video.
The maximum amount of text (measured in tokens) an LLM can process in a single interaction - both input and output combined.
A large, structured collection of text data used to train or evaluate natural language processing models.
AI that models cause-and-effect relationships rather than just statistical correlations, enabling more robust and explainable decision-making.
The process of splitting large documents into smaller segments before embedding them into a vector database for retrieval in RAG pipelines.
An alignment technique developed by Anthropic that trains models to follow a set of principles and self-critique their outputs for harmlessness.
A self-supervised learning approach where a model learns representations by distinguishing between similar and dissimilar data pairs.
AI systems designed for natural back-and-forth dialogue with users, combining NLU, dialogue management, and NLG components.
A loss function commonly used in classification and language modelling that measures the difference between predicted probability distributions and true labels.
A training strategy where a model is exposed to easier examples first and gradually introduced to harder ones, mimicking how humans learn.
