Large Language Models
328 views
Alignment Tax
Quick Definition
Performance cost of aligning AI with human preferences
Full Definition
The performance cost incurred when aligning AI models to follow human preferences and safety guidelines.
Examples
capability reduction after RLHF, safety vs helpfulness tradeoff
Related Terms
rlhf
ai-alignment
fine-tuning
More Large Language Models Terms
LoRA
Parameter-efficient fine-tuning using low-rank matrix decomposition
QLoRA
Combining 4-bit quantization with LoRA for efficiency
Prompt Template
Reusable structured format with dynamic input placeholders
BPE
Subword tokenization algorithm merging frequent character pairs
Instruction Tuning
Fine-tuning on instruction-response pairs for better following
PEFT
Fine-tuning methods updating only small parameter subsets