# Dynamic Proxy-Mixing: Transferring Replay Controllers from Small to Large Models for Continual Instruction Tuning

arXiv: https://arxiv.org/abs/2606.00400 · HTML: https://mlsanity.com/paper/2606.00400/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 15 of its 23 references that are in the map.
- First posted on arXiv 29 May 2026.

## Builds on (most cited first, 15 of 15)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Overcoming catastrophic forgetting in neural networks](https://mlsanity.com/paper/1612.00796/) (2016)
- [iCaRL: Incremental Classifier and Representation Learning](https://mlsanity.com/paper/1611.07725/) (2016)
- [Gradient Episodic Memory for Continual Learning](https://mlsanity.com/paper/1706.08840/) (2017)
- [Pythia: A Suite for Analyzing Large Language Models Across Training and Scaling](https://mlsanity.com/paper/2304.01373/) (2023)
- [Progressive Neural Networks](https://mlsanity.com/paper/1606.04671/) (2016)
- [Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!](https://mlsanity.com/paper/2310.03693/) (2023)
- [DoReMi: Optimizing Data Mixtures Speeds Up Language Model Pretraining](https://mlsanity.com/paper/2305.10429/) (2023)
- [Data Selection for Language Models via Importance Resampling](https://mlsanity.com/paper/2302.03169/) (2023)
- [Orthogonal Subspace Learning for Language Model Continual Learning](https://mlsanity.com/paper/2310.14152/) (2023)
- [Tensor Programs V: Tuning Large Neural Networks via Zero-Shot Hyperparameter Transfer](https://mlsanity.com/paper/2203.03466/) (2022)
- [Small-scale proxies for large-scale Transformer training instabilities](https://mlsanity.com/paper/2309.14322/) (2023)
- [MoDS: Model-oriented Data Selection for Instruction Tuning](https://mlsanity.com/paper/2311.15653/) (2023)
- [InsCL: A Data-efficient Continual Learning Paradigm for Fine-tuning Large Language Models with Instructions](https://mlsanity.com/paper/2403.11435/) (2024)
- [Data Mixing Agent: Learning to Re-weight Domains for Continual Pre-training](https://mlsanity.com/paper/2507.15640/) (2025)

## Built on by (most cited first, 3 of 3)

- [Rethinking Transfer in Continual Learning: A Replay-Based Realisation](https://mlsanity.com/paper/2607.15587/) (2026)
- [Small Models Scout Bottleneck Order for Large-Model Data Control](https://mlsanity.com/paper/2608.14936/) (2026)
- [A helps B while B hurts A: directed transfer in instruction-tuning mixture](https://mlsanity.com/paper/2609.39702/) (2026)

## Abstract (arXiv)

Continual instruction tuning updates a language model through a sequence of new domains, yet each update can progressively erode previously learned capabilities and alignment behavior. Replay is the standard mitigation, but fixed replay ratios are inherently limited because the optimal mixture varies with the current domain, the training stage, and the evolving vulnerability of prior behaviors. We propose PROX-YMIX, a framework that learns a dynamic replay controller on a small proxy model and transfers the frozen controller to a larger target. The controller never observes future tasks and constructs its state from normalized validation losses and their temporal dynamics, producing a masked mixture over the current task and accessible replay buffers. Our core empirical hypothesis is forgetting mirroring: task vulnerability rankings remain largely consistent across model scales even when absolute loss magnitudes differ. We validate this assumption empirically before transferring controllers across scales. On LLaMA-3-8B across five continual instruction tuning sequences, PROXYMIX improves average accuracy by 3.4 points, reduces final forgetting by 3.5 points, and raises safety score by 5.8 points over the strongest non-oracle baseline, at roughly 50x lower policy learning cost than Oracle Target RL. The framework is leakage free and architecture independent at the interface level, and we also identify settings where the proxy assumption breaks down, highlighting limitations for robust deployment.