The reviewed record of science sign in
Pith

arxiv: 2404.02657 · v4 · pith:AQBO3LKX · submitted 2024-04-03 · cs.CL · cs.AI

Rethinking Kullback-Leibler Divergence in Knowledge Distillation for Large Language Models

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:AQBO3LKXrecord.jsonopen to challenge →

classification cs.CL cs.AI
keywords divergenceepochskullback-leiblerllmsdistillationfocusesgithubknowledge
0
0 comments X
read the original abstract

Kullback-Leiber divergence has been widely used in Knowledge Distillation (KD) to compress Large Language Models (LLMs). Contrary to prior assertions that reverse Kullback-Leibler (RKL) divergence is mode-seeking and thus preferable over the mean-seeking forward Kullback-Leibler (FKL) divergence, this study empirically and theoretically demonstrates that neither mode-seeking nor mean-seeking properties manifest in KD for LLMs. Instead, RKL and FKL are found to share the same optimization objective and both converge after a sufficient number of epochs. However, due to practical constraints, LLMs are seldom trained for such an extensive number of epochs. Meanwhile, we further find that RKL focuses on the tail part of the distributions, while FKL focuses on the head part at the beginning epochs. Consequently, we propose a simple yet effective Adaptive Kullback-Leiber (AKL) divergence method, which adaptively allocates weights to combine FKL and RKL. Metric-based and GPT-4-based evaluations demonstrate that the proposed AKL outperforms the baselines across various tasks and improves the diversity and quality of generated responses. Codes are available at \href{https://github.com/wutaiqiang/LLM_KD_AKL}{github}.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Weak-to-Strong Generalization via Direct On-Policy Distillation

    cs.LG 2026-07 conditional novelty 7.0

    Transferring the log-ratio of a small model's pre-RL and post-RL checkpoints provides a dense implicit reward that improves stronger student models at a fraction of the cost of direct RL.

  2. Measuring Model Robustness via Fisher Information: Spectral Bounds, Theoretical Guarantees, and Practical Algorithms

    cs.LG 2026-06 unverdicted novelty 6.0

    Proposes spectral norm of Fisher Information Matrix as attack-agnostic robustness metric with closed-form bounds for common architectures and correlation to adversarial vulnerability.

  3. Contribution Weights: A Geometrical Analysis of Self-Attention Transformers

    cs.LG 2026-05 unverdicted novelty 6.0

    Contribution Weights combine attention, value magnitude, and directional alignment to measure token influence more faithfully than attention alone, and show attention sinks actively suppress information via a convex s...

  4. ADWIN: Adaptive Windows for Horizon-Aware On-Policy Distillation

    cs.LG 2026-05 unverdicted novelty 6.0

    ADWIN adaptively selects training horizons in on-policy distillation via prefix alignment checks, cutting end-to-end cost by up to 4.1x while matching or exceeding full-rollout accuracy on math and code benchmarks.

  5. When Are Teacher Tokens Reliable? Position-Weighted On-Policy Self-Distillation for Reasoning

    cs.LG 2026-05 unverdicted novelty 6.0

    Position-Weighted On-Policy Self-Distillation (PW-OPSD) weights later tokens more heavily after a diagnostic shows position predicts teacher reliability better than entropy, yielding +1.0 and +1.1 Avg@12 gains on AIME...