Enregistré dans:
Détails bibliographiques
Auteurs principaux: Subramanian, Shreyas, Krishnamoorthy, Bala, Murthy, Pranav
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2512.14527
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866915679244910592
author Subramanian, Shreyas
Krishnamoorthy, Bala
Murthy, Pranav
author_facet Subramanian, Shreyas
Krishnamoorthy, Bala
Murthy, Pranav
contents Despite significant advances in optimizers for training, most research works use common scheduler choices like Cosine or exponential decay. In this paper, we study \emph{GreedyLR}, a novel scheduler that adaptively adjusts the learning rate during training based on the current loss. To validate the effectiveness of our proposed scheduler, we conduct experiments on several NLP, CV, and LLM tasks with up to $7B$ parameters, including both fine-tuning and pre-training experiments. The results show that our approach outperforms several state-of-the-art schedulers in terms of accuracy, speed, and convergence. We also provide a theoretical analysis of the GreedyLR algorithm, including a proof of convergence and derivation of the optimal scaling factor $F$ that maximizes the convergence rate, along with experiments to show robustness of the algorithm to realistic noisy landscapes. Our scheduler is easy to implement, computationally efficient, and could be considered a good default scheduler for training.
format Preprint
id arxiv_https___arxiv_org_abs_2512_14527
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Dynamic Learning Rate Scheduling based on Loss Changes Leads to Faster Convergence
Subramanian, Shreyas
Krishnamoorthy, Bala
Murthy, Pranav
Artificial Intelligence
Despite significant advances in optimizers for training, most research works use common scheduler choices like Cosine or exponential decay. In this paper, we study \emph{GreedyLR}, a novel scheduler that adaptively adjusts the learning rate during training based on the current loss. To validate the effectiveness of our proposed scheduler, we conduct experiments on several NLP, CV, and LLM tasks with up to $7B$ parameters, including both fine-tuning and pre-training experiments. The results show that our approach outperforms several state-of-the-art schedulers in terms of accuracy, speed, and convergence. We also provide a theoretical analysis of the GreedyLR algorithm, including a proof of convergence and derivation of the optimal scaling factor $F$ that maximizes the convergence rate, along with experiments to show robustness of the algorithm to realistic noisy landscapes. Our scheduler is easy to implement, computationally efficient, and could be considered a good default scheduler for training.
title Dynamic Learning Rate Scheduling based on Loss Changes Leads to Faster Convergence
topic Artificial Intelligence
url https://arxiv.org/abs/2512.14527