DistiLLM: Towards Streamlined Distillation for Large Language Models

Code

• github.com/jongwooko/distillm
  OfficialIn paperjax★ 255
• github.com/tianyic/only_train_once
  pytorch★ 311
• github.com/jongwooko/distillm-2
  pytorch★ 59
• github.com/ghwang-s/abkd
  pytorch★ 44
• github.com/mingyukim87/synergynerf
  pytorch★ 10

Abstract

Knowledge distillation (KD) is widely used for compressing a teacher model to a smaller student model, reducing its inference cost and memory footprint while preserving model capabilities. However, current KD methods for auto-regressive sequence models (e.g., large language models) suffer from missing a standardized objective function. Moreover, the recent use of student-generated outputs to address training-inference mismatches has significantly escalated computational costs. To tackle these issues, we introduce DistiLLM, a more effective and efficient KD framework for auto-regressive language models. DistiLLM comprises two components: (1) a novel skew Kullback-Leibler divergence loss, where we unveil and leverage its theoretical properties, and (2) an adaptive off-policy approach designed to enhance the efficiency in utilizing student-generated outputs. Extensive experiments, including instruction-following tasks, demonstrate the effectiveness of DistiLLM in building high-performing student models while achieving up to 4.3 speedup compared to recent KD methods.

Tasks

Reproductions