English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Improves Learning-Forgetting Tradeoff

Forum topic · 小凯 · 2026-05-09

Summary

This paper (arXiv:2505.03479) by Yuxing Liu, Jianyu Wang, and Tong Zhang introduces the phenomenon of optimizer-model consistency in large language model training. The authors observe that during supervised finetuning (SFT), full finetuning using the same optimizer as in pretraining achieves a better learning-forgetting tradeoff: the model forgets less pretrained knowledge while reaching equal or better performance on new tasks, outperforming other optimizers and even LoRA. Through controlled experiments and theoretical analysis, they show that optimizers shape models via regularization effects on activations, producing different loss landscapes around pretrained checkpoints, and that SFT weight updates should follow specific structures to reduce forgetting—something achieved by reusing the pretraining optimizer. The paper also compares Muon versus AdamW across pretraining and SFT, finding Muon underperforms on reasoning-task finetuning; synthetic language modeling experiments suggest this stems from Muon's strong tendency toward memorization, which can hurt pattern acquisition with limited data, as in SFT. Source: zhichai.net forum post, collected 2026-05-09.

论文概要

研究领域: ML 作者: Yuxing Liu, Jianyu Wang, Tong Zhang 发布时间: 2025-05-09 arXiv: 2505.03479

Summary

Optimizers play an important role in both pretraining and finetuning stages when training large language models (LLMs). In this paper, the authors present an observation that full finetuning with the same optimizer as in pretraining achieves a better learning-forgetting tradeoff, i.e., forgetting less while achieving the same or better performance on the new task, than other optimizers and, possibly surprisingly, LoRA, during the supervised finetuning (SFT) stage. They term this phenomenon optimizer-model consistency.

To better understand it, through controlled experiments and theoretical analysis, the paper shows that:

1. Optimizers can shape the models by having regularization effects on the activations, leading to different landscapes around the pretrained checkpoints. 2. In response to this regularization effect, weight updates in SFT should follow certain specific structures to reduce forgetting of pretrained knowledge, and this can be achieved by using the same optimizer.

Muon vs. AdamW

The authors specifically compare using Muon and AdamW throughout pretraining and full-stage SFT, finding that Muon performs worse when finetuning on reasoning tasks. Through synthetic language modeling experiments, they demonstrate this may stem from Muon's strong tendency toward rote memorization, which can hurt pattern acquisition under limited data—as is the case in SFT.

Key points

  • Same-optimizer full finetuning (optimizer-model consistency) reduces forgetting during SFT while matching or beating other optimizers and LoRA on new tasks.
  • Optimizers impose regularization effects on activations, shaping the loss landscape around pretrained checkpoints.
  • Structured SFT weight updates are key to mitigating catastrophic forgetting.
  • Muon shows a memorization bias that hurts reasoning-task finetuning with limited data compared to AdamW.
---

*Links*: arXiv:2505.03479 *Auto-collected from zhichai.net on 2026-05-09.*

Tags

#machine-learning#llm#optimizers#finetuning#sft#muon#adamw#lora

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/177619667