MiniDisc: Minimal Distillation Schedule for Language Model Compression
Recent studies have uncovered that language model distillation is less effective when facing a large capacity gap between the teacher and the student, and introduced teacher assistant-based distillation to bridge the gap. As a connection, the scale and the performance of the teacher assistant is of vital importance to bring the knowledge from the teacher to the student. However, existing teacher assistant-based methods require maximally many trials before scheduling an optimal teacher assistant. To this end, we propose a minimal distillation schedule (MiniDisc) for scheduling the optimal teacher assistant in minimally one trial. In particular, motivated by the finding that the performance of the student is positively correlated to the scale-performance tradeoff of the teacher assistant, MiniDisc is designed with a $\lambda$-tradeoff to measure the optimality of the teacher assistant without trial distillation to the student. MiniDisc then can schedule the optimal teacher assistant with the best $\lambda$-tradeoff in a sandwich framework. MiniDisc is evaluated with an extensive set of experiments on GLUE. Experimental results demonstrate the improved efficiency our MiniDisc compared to several state-of-the-art baselines. We further apply MiniDisc to a language model with billions of parameters and show its scalability.
Code (1)
Tasks
Knowledge DistillationLanguage ModelingLanguage ModellingModel CompressionSchedulingMethods 이 논문이 사용한 방법론
Similar Papers 제목 키워드 기반
Scaling Laws for Task-Specific LLM Distillation
Large Language Models (LLMs) achieve strong performance across a growing range of domains, yet their scale poses deployment challenges in applications where latency and cost constraints are critical. This paper derives e…
General KnowledgeSlimQwen: Exploring the Pruning and Distillation in Large MoE Model Pre-training
Structured pruning and knowledge distillation (KD) are typical techniques for compressing large language models, but it remains unclear how they should be applied at pretraining scale, especially to recent mixture-of-exp…
Knowledge DistillationContinual PretrainingConstrained Edge AI Deployment: Fine-Tuning vs Distillation for LLM Compression
Modern foundational models are often compressed via a combination of structured pruning and re-training to meet the strict compute, memory, and connectivity constraints of edge deployments. While state-of-the-art pruning…
PISCO: Pretty Simple Compression for Retrieval-Augmented Generation
Retrieval-Augmented Generation (RAG) pipelines enhance Large Language Models (LLMs) by retrieving relevant documents, but they face scalability issues due to high inference costs and limited context size. Document compre…
GPUKnowledge DistillationQuestion AnsweringRAG+2X-AuT: Progressive Audio-Encoder Compression for Speech LLMs with Cross-Scale Distillation
Reducing audio-encoder depth lowers the inference cost of speech large language models, but removing complete blocks perturbs the embeddings consumed by the decoder and can cause deletion and premature end-of-sequence er…