Breaking the Tokenizer Barrier: On-Policy Distillation across Model Families

• URL: http://arxiv.org/abs/2606.09456v1
• Date: Mon, 08 Jun 2026 13:12:01 GMT
• Title: Breaking the Tokenizer Barrier: On-Policy Distillation across Model Families
• Abstract summary: On-Policy Distillation (OPD) has become a core technique in the post-training of Large Language Models (LLMs) for transferring knowledge from domain experts to student models.<n>Existing OPD distillation methods require teacher and student models to share the same tokenizer, restricting the applicability of OPD within the model series.<n>In this work, we enable the standard on-policy distillation method to operate across model families, ensuring that high-fidelity token-level signals can propagate across different tokenizers.
• Score: 32.34854415784466
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: On-Policy Distillation (OPD) has become a core technique in the post-training of Large Language Models (LLMs) for transferring knowledge from domain experts to student models. However, existing OPD distillation methods require teacher and student models to share the same tokenizer, restricting the applicability of OPD within the model series. Current mainstream practice typically employs Supervised Fine-Tuning (SFT) on teacher-generated responses for cross-tokenizer distillation, which fails to capture the rich knowledge embedded in the teacher's probability distribution. In this work, we enable the standard on-policy distillation method to operate across model families, ensuring that high-fidelity token-level signals can propagate across different tokenizers with a precise token-mapping algorithm. Extensive experiments show that cross-tokenizer OPD is significantly more compute-efficient than baselines on various benchmarks. Our results unlock a broader range of teacher-student pairs for OPD, opening up new avenues for adapting and enhancing interactions between LLMs.

Related papers

• On-Policy Context Distillation for Language Models [92.82835176360864]
  We propose On-Policy Context Distillation (OPCD), a framework that bridges on-policy distillation with context distillation.<n>We demonstrate the effectiveness of OPCD on two important applications: experiential knowledge distillation and system prompt distillation.
  arXiv  Detail & Related papers  (2026-02-12T18:58:28Z)
• OVD: On-policy Verbal Distillation [47.727229201069555]
  On-policy Verbal Distillation (OVD) is a memory-efficient framework that replaces token-level probability matching with trajectory matching.<n>OVD dramatically reduces memory consumption while enabling on-policy distillation from teacher models with verbal feedback.
  arXiv  Detail & Related papers  (2026-01-29T16:48:14Z)
• CTPD: Cross Tokenizer Preference Distillation [19.4149691480574]
  Cross-Tokenizer Preference Distillation (CTPD) is the first unified framework for transferring human-aligned behavior between models with heterogeneous tokenizers.<n>CTPD introduces three key innovations: (1) Aligned Span Projection, which maps teacher and student to shared character-level spans for precise supervision transfer; (2) a cross-tokenizer adaptation of Token-level Importance Sampling (TIS-DPO) for improved credit assignment; and (3) a Teacher-Anchored Reference, allowing the student to directly leverage the teacher's preferences in a DPO-style objective.
  arXiv  Detail & Related papers  (2026-01-17T01:11:35Z)
• Intra-class Patch Swap for Self-Distillation [3.282914142012984]
  We propose a teacher-free distillation framework based on a single student network.<n>Our approach is built on a simple yet highly effective augmentation, called intra-class patch swap augmentation.<n>Our method consistently outperforms both existing self-distillation baselines and conventional teacher-based KD approaches.
  arXiv  Detail & Related papers  (2025-05-20T09:30:19Z)
• Learning from Stochastic Teacher Representations Using Student-Guided Knowledge Distillation [64.15918654558816]
  Self-distillation (SSD) training strategy is introduced for filtering and weighting teacher representation to distill from task-relevant representations only.<n> Experimental results on real-world affective computing, wearable/biosignal datasets from the UCR Archive, the HAR dataset, and image classification datasets show that the proposed SSD method can outperform state-of-the-art methods.
  arXiv  Detail & Related papers  (2025-04-19T14:08:56Z)
• Universal Cross-Tokenizer Distillation via Approximate Likelihood Matching [16.385782508179364]
  Distillation has shown remarkable success in transferring knowledge from a Large Language Model (LLM) teacher to a student LLM.<n>Current distillation methods require similar tokenizers between the teacher and the student, restricting their applicability to only a small subset of teacher-student pairs.<n>We develop a principled cross-tokenizer distillation method to solve this crucial deficiency.<n>Our method is the first to enable effective distillation across fundamentally different tokenizers, while also substantially outperforming prior methods in all other cases.
  arXiv  Detail & Related papers  (2025-03-25T21:44:10Z)
• Exploring and Enhancing the Transfer of Distribution in Knowledge Distillation for Autoregressive Language Models [62.5501109475725]
  Knowledge distillation (KD) is a technique that compresses large teacher models by training smaller student models to mimic them. This paper introduces Online Knowledge Distillation (OKD), where the teacher network integrates small online modules to concurrently train with the student model. OKD achieves or exceeds the performance of leading methods in various model architectures and sizes, reducing training time by up to fourfold.
  arXiv  Detail & Related papers  (2024-09-19T07:05:26Z)
• Multi Teacher Privileged Knowledge Distillation for Multimodal Expression Recognition [58.41784639847413]
  Human emotion is a complex phenomenon conveyed and perceived through facial expressions, vocal tones, body language, and physiological signals. In this paper, a multi-teacher PKD (MT-PKDOT) method with self-distillation is introduced to align diverse teacher representations before distilling them to the student. Results indicate that our proposed method can outperform SOTA PKD methods.
  arXiv  Detail & Related papers  (2024-08-16T22:11:01Z)
• Multi-Granularity Semantic Revision for Large Language Model Distillation [66.03746866578274]
  We propose a multi-granularity semantic revision method for LLM distillation. At the sequence level, we propose a sequence correction and re-generation strategy. At the token level, we design a distribution adaptive clipping Kullback-Leibler loss as the distillation objective function. At the span level, we leverage the span priors of a sequence to compute the probability correlations within spans, and constrain the teacher and student's probability correlations to be consistent.
  arXiv  Detail & Related papers  (2024-07-14T03:51:49Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.