My custom AI: LLM industry aligned taxonomy

Style Attribute Control for Controllability in LLM Outputs: In-world Artifacts that describe attributes like flourish, rhetorical, stylistic, performative, human-like interaction etc.

Topology Taxonomy of Reasoning Schemes -> Reasoning & Deductive Benchmarks: In-world Artifacts that describe Logic seeds, Mathematical Reasoning, Chain of Thought Validity, Deductive/Inductive evaluation etc.

Emergent Abilities (Scaling-Laws Literature); abilities that appear suddenly with model scale, not explicitly trained: In-world Artifacts that give Emergent Logic Skills and Emergent Flourishey Performative Skills.

LLM Evaluation Taxonomy -> Epistemology sub-domain -> philosophical roots :: HELM, MMLU, and Anthropic HH benchmarks: In-world Artifacts that guide epistemology and philosophy

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.