Large Language Models
142 views
Guardrails LLM
Quick Definition
Frameworks for monitoring and controlling LLM I/O
Full Definition
Frameworks for monitoring and controlling LLM inputs and outputs in production.
Examples
output validation, content filtering, compliance checking
Related Terms
ai-safety
content-moderation
More Large Language Models Terms
Semantic Search
Search technique understanding meaning rather than keywords
Adapter Layer
Small trainable modules in frozen transformer layers
GPT
OpenAI's family of autoregressive large language models
RLHF
Aligning LLMs with human preferences using reinforcement learning
Alignment Tax
Performance cost of aligning AI with human preferences
Prompt Caching
Storing key-value states for repeated prompt prefixes