Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Chain-based Distillation for Effective Initialization of Variable-Sized Small Language Models

About

Large language models (LLMs) achieve strong performance but remain costly to deploy in resource-constrained settings. Training small language models (SLMs) from scratch is computationally expensive, while conventional knowledge distillation requires repeated access to large teachers for different target sizes, leading to poor scalability. To solve these problems, we propose \textbf{Chain-based Distillation (CBD)}, a scalable paradigm for efficiently initializing variable-sized language models. A sparse and limited sequence of intermediate models (called anchors) is constructed via stepwise distillation, forming a distillation chain that progressively transfers knowledge from the source LLMs. To support heterogeneous settings, we introduce \emph{bridge distillation} for cross-architecture and cross-vocabulary transfer. Models of variable sizes are initialized via parameter interpolation between adjacent anchors, eliminating repeated large teacher inference. Experiments show that the proposed method substantially improves efficiency and downstream performance. A 138M-parameter SLM without recovery pre-training, outperforms scratch-trained models on a 10B-token corpus on the specific task. CBD also demonstrates versatility in heterogeneous settings for initialize models with different architectures and vocabularies.

Boyu Shi, YiCheng Jiang, Chang Liu, Qiufeng Wang, Xu Yang, Xin Geng• 2026

Related benchmarks

TaskDatasetResultRank
Commonsense ReasoningWinoGrande
Accuracy62
1581
Sentence CompletionHellaSwag
Accuracy43.03
440
Multiple-choice Question AnsweringMMLU
Accuracy31
222
Instruction FollowingDolly
Score22.99
36
Downstream Performance EvaluationDownstream Tasks BoolQ, MMLU, WinoG
BoolQ Accuracy71.9
5
Multiple-choice Question AnsweringTinyARC
Accuracy29.5
2
Multiple-choice Question AnsweringTinyMMLU
Accuracy28.2
2
Showing 7 of 7 rows

Other info

Follow for update