The reviewed record of science sign in
Pith

arxiv: 2502.08606 · v2 · pith:IJOWF3PN · submitted 2025-02-12 · cs.LG · cs.AI· cs.CL· stat.ML

Distillation Scaling Laws

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:IJOWF3PNrecord.jsonopen to challenge →

classification cs.LG cs.AIcs.CLstat.ML
keywords distillationteacherstudentallocationcomputecompute-optimaldistilledlarge-scale
0
0 comments X
read the original abstract

We propose a distillation scaling law that estimates distilled model performance based on a compute budget and its allocation between the student and teacher. Our findings mitigate the risks associated with large-scale distillation by enabling compute-optimal allocation for both the teacher and student to maximize student performance. We provide compute-optimal distillation recipes for two key scenarios: when a teacher already exists, and when a teacher needs training. In settings involving many students or an existing teacher, distillation outperforms supervised learning up to a compute level that scales predictably with student size. Conversely, if only one student is to be distilled and a teacher also requires training, supervised learning is generally preferable. Additionally, our large-scale study of distillation increases our understanding of the process and helps inform experimental design.

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 11 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. Unmasking On-Policy Distillation: Where It Helps, Where It Hurts, and Why

    cs.LG 2026-05 unverdicted novelty 7.0

    Distillation signals align better with ideal updates on incorrect student rollouts than correct ones, with optimal teacher context depending on student capacity and task.

  3. Locking Pretrained Weights via Deep Low-Rank Residual Distillation

    cs.LG 2026-05 unverdicted novelty 7.0

    DLR-Lock locks open-weight LLMs against unauthorized fine-tuning by swapping MLPs for deep low-rank residual networks that inflate backprop memory and complicate optimization, yet preserve original capabilities via mo...

  4. When Does Generating More Help? Disentangling Fixed-Source Synthesis from Source Expansion in Synthetic Data Scaling

    cs.CL 2026-07 unverdicted novelty 6.0

    Fixed-source synthesis is bounded; a derived scaling law predicts high-budget performance from low-budget fits, and source expansion outperforms fixed-source at large matched budgets.

  5. Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe

    cs.LG 2026-04 unverdicted novelty 6.0

    On-policy distillation works when student and teacher models share thinking patterns and the teacher adds new capabilities, with success tied to alignment on a small set of high-probability tokens.

  6. Attention to Mamba: A Recipe for Cross-Architecture Distillation

    cs.CL 2026-04 unverdicted novelty 6.0

    A two-stage distillation recipe converts a Pythia-1B Transformer into a Mamba model that preserves performance with perplexity 14.11 versus the teacher's 13.86.

  7. DOPD: Dual On-policy Distillation

    cs.AI 2026-06 unverdicted novelty 5.0

    DOPD is an advantage-aware dual distillation method that dynamically assigns token supervision from either privileged teacher or student to transfer capability while mitigating non-replicable information asymmetry in ...

  8. Scaling Laws for Task-Specific LLM Distillation

    cs.AI 2026-06 unverdicted novelty 5.0

    Empirical scaling laws for task-specific LLM distillation in quantitative finance indicate that chain-of-thought supervision recovers general knowledge lost during iterative pruning while in-domain performance degrade...

  9. GAMMA: Global Bit Allocation for Mixed-Precision Models under Arbitrary Budgets

    cs.LG 2026-05 unverdicted novelty 5.0

    GAMMA is a post-training framework that learns stable module sensitivity rankings for mixed-precision LLM quantization and projects them to exact bit budgets via integer programming, enabling reuse across arbitrary me...

  10. Towards Reasoning Era: A Survey of Long Chain-of-Thought for Reasoning Large Language Models

    cs.AI 2025-03 unverdicted novelty 5.0

    The paper unifies perspectives on Long CoT in reasoning LLMs by introducing a taxonomy, detailing characteristics of deep reasoning and reflection, and discussing emergence phenomena and future directions.

  11. A Primer in Post-Training Reasoning Data: What We Know About How It Works

    cs.CL 2026-06 unverdicted novelty 2.0

    A literature synthesis that organizes post-training reasoning data research around data objects, usefulness factors, construction methods, and scaling behaviors to create an attribution framework.