# The Warmup Dilemma: How Learning Rate Strategies Impact Speech-to-Text Model Convergence

arXiv: https://arxiv.org/abs/2505.23420 · HTML: https://mlsanity.com/paper/2505.23420/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 21 of its 33 references that are in the map.
- First posted on arXiv 29 May 2025.

## Builds on (most cited first, 21 of 21)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [Rethinking the Inception Architecture for Computer Vision](https://mlsanity.com/paper/1512.00567/) (2015)
- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [Conformer: Convolution-augmented Transformer for Speech Recognition](https://mlsanity.com/paper/2005.08100/) (2020)
- [Common Voice: A Massively-Multilingual Speech Corpus](https://mlsanity.com/paper/1912.06670/) (2019)
- [Subword Regularization: Improving Neural Network Translation Models with Multiple Subword Candidates](https://mlsanity.com/paper/1804.10959/) (2018)
- [MLS: A Large-Scale Multilingual Dataset for Speech Research](https://mlsanity.com/paper/2012.03411/) (2020)
- [FLEURS: Few-shot Learning Evaluation of Universal Representations of Speech](https://mlsanity.com/paper/2205.12446/) (2022)
- [Libri-Light: A Benchmark for ASR with Limited or No Supervision](https://mlsanity.com/paper/1912.07875/) (2019)
- [VoxPopuli: A Large-Scale Multilingual Speech Corpus for Representation Learning, Semi-Supervised Learning and Interpretation](https://mlsanity.com/paper/2101.00390/) (2021)
- [Google USM: Scaling Automatic Speech Recognition Beyond 100 Languages](https://mlsanity.com/paper/2303.01037/) (2023)
- [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://mlsanity.com/paper/2010.05171/) (2020)
- [Branchformer: Parallel MLP-Attention Architectures to Capture Local and Global Context for Speech Recognition and Understanding](https://mlsanity.com/paper/2207.02971/) (2022)
- [Training Tips for the Transformer Model](https://mlsanity.com/paper/1804.00247/) (2018)
- [OWSM v3.1: Better and Faster Open Whisper-Style Speech Models based on E-Branchformer](https://mlsanity.com/paper/2401.16658/) (2024)
- [Why Warmup the Learning Rate? Underlying Mechanisms and Improvements](https://mlsanity.com/paper/2406.09405/) (2024)
- [A Comparative Study on End-to-end Speech to Text Translation](https://mlsanity.com/paper/1911.08870/) (2019)
- [The Model Openness Framework: Promoting Completeness and Openness for Reproducibility, Transparency, and Usability in Artificial Intelligence](https://mlsanity.com/paper/2403.13784/) (2024)
- [When Good and Reproducible Results are a Giant with Feet of Clay: The Importance of Software Quality in NLP](https://mlsanity.com/paper/2303.16166/) (2023)
- [MOSEL: 950,000 Hours of Speech Data for Open-Source Speech Foundation Model Training on EU Languages](https://mlsanity.com/paper/2410.01036/) (2024)

## Built on by (most cited first, 1 of 1)

- [Cross-Attention is Half Explanation in Speech-to-Text Models](https://mlsanity.com/paper/2509.18010/) (2025)

## Abstract (arXiv)

Training large-scale models presents challenges not only in terms of resource requirements but also in terms of their convergence. For this reason, the learning rate (LR) is often decreased when the size of a model is increased. Such a simple solution is not enough in the case of speech-to-text (S2T) trainings, where evolved and more complex variants of the Transformer architecture -- e.g., Conformer or Branchformer -- are used in light of their better performance. As a workaround, OWSM designed a double linear warmup of the LR, increasing it to a very small value in the first phase before updating it to a higher value in the second phase. While this solution worked well in practice, it was not compared with alternative solutions, nor was the impact on the final performance of different LR warmup schedules studied. This paper fills this gap, revealing that i) large-scale S2T trainings demand a sub-exponential LR warmup, and ii) a higher LR in the warmup phase accelerates initial convergence, but it does not boost final performance.