Full search page →

Discriminative learning rates and gradual unfreezing

Optional · supplementary note 4 of 4 for DLCV 01A · Training Deep Vision Models in Practice. Nothing in the main text depends on it; citation numbers refer to the reference list at the bottom of this page.

Navigate: this page and this track: Navigate: this page and this track

On this page

Supplementary notes · 4

  1. Cosine annealing with warm restarts
  2. Gradient accumulation when the loss is not a plain mean
  3. EMA of weights: choosing the decay
  4. Discriminative learning rates and gradual unfreezing

Paper guides · 2

  1. Accurate, Large Minibatch SGD (Goyal et al.)
  2. Fine-Tuning can Distort Pretrained Features (Kumar et al.)

Deep Learning for Computer Vision

  1. 02 · Deep Image Classification
View the whole course tree →

This note shows how to set a separate learning rate for each layer group and how to unfreeze a pretrained network in stages.

Discriminative learning rates. Split the parameters into groups θ1,…,θL\theta^1, \dots, \theta^L from the input (l=1l = 1) to the head (l=Ll = L), and give group ll its own rate ηl\eta^l. The update is θtl=θt−1l−ηl ∇θlJ\theta^l_t = \theta^l_{t-1} - \eta^l\,\nabla_{\theta^l} J, where JJ is the training objective. Howard and Ruder chose the head’s rate ηL\eta^L by fine-tuning only the last layer and then set ηl−1=ηl/2.6\eta^{l-1} = \eta^l / 2.6 for each lower layer [34]; the factor 2.6 is an empirical choice from their language-model experiments, not a universal constant. The general pattern is a geometric decay,

ηl=ηL α L−l,0<α≤1,\eta^l = \eta^L\,\alpha^{\,L - l}, \qquad 0 \lt \alpha \le 1,

where α\alpha is the per-group decay factor: α=1\alpha = 1 is ordinary fine-tuning, and small α\alpha approaches freezing the early layers. In practice, use groups such as the stem, each backbone stage and the head, rather than every layer.

Gradual unfreezing. Instead of training everything from the start, unfreeze the last layer group first and train for one epoch, then unfreeze the next lower group and train all unfrozen groups for another epoch, and continue until the whole network trains [34]. The motivation is the generality gradient measured by Yosinski et al.: later layers are the most task-specific and should adapt first [32].

Edge cases. (1) Frozen layers that contain BN should normally be in evaluation mode too, otherwise their running statistics keep changing even though their weights do not. (2) Keep frozen parameters out of the optimizer, or give them no gradient, so that no update of any kind, including weight decay, touches them. (3) When a group is unfrozen, its optimizer state (momentum, or Adam’s moment estimates) starts from zero; a short warmup for that group avoids a sudden jump. (4) A schedule defined as a multiplier of each group’s base rate keeps the ratios ηl/ηL\eta^l / \eta^L fixed during warmup and decay.

References

  1. T. He, Z. Zhang, H. Zhang, Z. Zhang, J. Xie and M. Li, “Bag of tricks for image classification with convolutional neural networks,” in Proc. IEEE/CVF CVPR, 2019. doi
  2. A. Karpathy, “A recipe for training neural networks,” blog post, 2019. link
  3. scikit-learn developers, “load_digits,” scikit-learn documentation, version 1.9.1. link
  4. M. Buda, A. Maki and M. A. Mazurowski, “A systematic study of the class imbalance problem in convolutional neural networks,” Neural Networks, vol. 106, 2018. doi
  5. T.-Y. Lin, P. Goyal, R. Girshick, K. He and P. Dollár, “Focal loss for dense object detection,” in Proc. IEEE ICCV, 2017. doi
  6. A. Buslaev, V. I. Iglovikov, E. Khvedchenya, A. Parinov, M. Druzhinin and A. A. Kalinin, “Albumentations: Fast and flexible image augmentations,” Information, vol. 11, no. 2, 2020. doi
  7. C. Shorten and T. M. Khoshgoftaar, “A survey on image data augmentation for deep learning,” Journal of Big Data, vol. 6, 2019. doi
  8. K. He, X. Zhang, S. Ren and J. Sun, “Delving deep into rectifiers: Surpassing human-level performance on ImageNet classification,” in Proc. IEEE ICCV, 2015. doi
  9. P. Goyal, P. Dollár, R. Girshick, P. Noordhuis, L. Wesolowski, A. Kyrola, A. Tulloch, Y. Jia and K. He, “Accurate, large minibatch SGD: Training ImageNet in 1 hour,” arXiv:1706.02677 (technical report), 2017. arXiv
  10. S. Ioffe and C. Szegedy, “Batch normalization: Accelerating deep network training by reducing internal covariate shift,” in Proc. ICML, PMLR 37, 2015. link
  11. Y. Wu and K. He, “Group normalization,” in Proc. ECCV, 2018. doi
  12. Y. Bengio, “Practical recommendations for gradient-based training of deep architectures,” in Neural Networks: Tricks of the Trade, 2nd ed., LNCS 7700, Springer, 2012. doi
  13. L. N. Smith, “Cyclical learning rates for training neural networks,” in Proc. IEEE WACV, 2017. doi
  14. L. N. Smith, “A disciplined approach to neural network hyper-parameters: Part 1 — learning rate, batch size, momentum, and weight decay,” US Naval Research Laboratory Technical Report 5510-026, arXiv:1803.09820, 2018. arXiv
  15. I. Loshchilov and F. Hutter, “SGDR: Stochastic gradient descent with warm restarts,” in Proc. ICLR, 2017. arXiv
  16. L. N. Smith and N. Topin, “Super-convergence: Very fast training of neural networks using large learning rates,” in Artificial Intelligence and Machine Learning for Multi-Domain Operations Applications, Proc. SPIE 11006, 2019. doi
  17. PyTorch contributors, “torch.optim.lr_scheduler.OneCycleLR,” PyTorch 2.15 documentation, 2026. link
  18. K. He, X. Zhang, S. Ren and J. Sun, “Deep residual learning for image recognition,” in Proc. IEEE CVPR, 2016. doi
  19. J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li and L. Fei-Fei, “ImageNet: A large-scale hierarchical image database,” in Proc. IEEE CVPR, 2009. doi
  20. D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” in Proc. ICLR, 2015. arXiv
  21. I. Loshchilov and F. Hutter, “Decoupled weight decay regularization,” in Proc. ICLR, 2019. arXiv
  22. PyTorch contributors, “torch.optim.Adam,” PyTorch 2.15 documentation, 2026. link
  23. N. Srivastava, G. Hinton, A. Krizhevsky, I. Sutskever and R. Salakhutdinov, “Dropout: A simple way to prevent neural networks from overfitting,” Journal of Machine Learning Research, vol. 15, 2014. link
  24. G. Huang, Y. Sun, Z. Liu, D. Sedra and K. Q. Weinberger, “Deep networks with stochastic depth,” in Proc. ECCV, 2016. doi
  25. L. Prechelt, “Early stopping — but when?,” in Neural Networks: Tricks of the Trade, 2nd ed., LNCS 7700, Springer, 2012. doi
  26. V. Godbole, G. E. Dahl, J. Gilmer, C. J. Shallue and Z. Nado, “Deep learning tuning playbook,” Google Research, GitHub repository, 2023. link
  27. B. T. Polyak and A. B. Juditsky, “Acceleration of stochastic approximation by averaging,” SIAM Journal on Control and Optimization, vol. 30, no. 4, 1992. doi
  28. P. Izmailov, D. Podoprikhin, T. Garipov, D. Vetrov and A. G. Wilson, “Averaging weights leads to wider optima and better generalization,” in Proc. UAI, 2018. arXiv
  29. D. Morales-Brotons, T. Vogels and H. Hendrikx, “Exponential moving average of weights in deep learning: Dynamics and benefits,” Transactions on Machine Learning Research, 2024. arXiv
  30. P. Micikevicius, S. Narang, J. Alben, G. Diamos, E. Elsen, D. Garcia, B. Ginsburg, M. Houston, O. Kuchaiev, G. Venkatesh and H. Wu, “Mixed precision training,” in Proc. ICLR, 2018. arXiv
  31. PyTorch contributors, “Automatic mixed precision package — torch.amp,” PyTorch 2.15 documentation, 2026. link
  32. J. Yosinski, J. Clune, Y. Bengio and H. Lipson, “How transferable are features in deep neural networks?,” in Advances in Neural Information Processing Systems 27, 2014. link
  33. E. J. Hu, Y. Shen, P. Wallis, Z. Allen-Zhu, Y. Li, S. Wang, L. Wang and W. Chen, “LoRA: Low-rank adaptation of large language models,” in Proc. ICLR, 2022. arXiv
  34. J. Howard and S. Ruder, “Universal language model fine-tuning for text classification,” in Proc. ACL, 2018. doi
  35. A. Kumar, A. Raghunathan, R. Jones, T. Ma and P. Liang, “Fine-tuning can distort pretrained features and underperform out-of-distribution,” in Proc. ICLR, 2022. arXiv
  36. C. Zhang, S. Bengio, M. Hardt, B. Recht and O. Vinyals, “Understanding deep learning requires rethinking generalization,” in Proc. ICLR, 2017. arXiv
  37. R. Geirhos, J.-H. Jacobsen, C. Michaelis, R. Zemel, W. Brendel, M. Bethge and F. A. Wichmann, “Shortcut learning in deep neural networks,” Nature Machine Intelligence, vol. 2, 2020. doi
  38. PyTorch contributors, “Reproducibility,” PyTorch 2.15 documentation, 2026. link
  39. Y. Xin, J. Yang, S. Luo, Y. Du, Q. Qin, K. Cen, Y. He, Z. Zhang, B. Fu, X. Yang, G. Zhai, M.-H. Yang and X. Liu, “Parameter-efficient fine-tuning for pre-trained vision models: A survey and benchmark,” arXiv:2402.02242, 2024. arXiv
  40. G. E. Dahl, F. Schneider, Z. Nado, N. Agarwal, C. S. Sastry, P. Hennig, S. Medapati, R. Eschenhagen, P. Kasimbeg, D. Suo et al., “Benchmarking neural network training algorithms,” arXiv:2306.07179, 2023. arXiv
  41. P. Kasimbeg, F. Schneider, R. Eschenhagen, J. Bae, C. S. Sastry, M. Saroufim, B. Feng, L. Wright, E. Z. Yang, Z. Nado, S. Medapati, P. Hennig, M. Rabbat and G. E. Dahl, “Accelerating neural network training: An analysis of the AlgoPerf competition,” in Proc. ICLR, 2025. arXiv
  42. A. Defazio, X. A. Yang, H. Mehta, K. Mishchenko, A. Khaled and A. Cutkosky, “The road less scheduled,” in Advances in Neural Information Processing Systems 37, 2024. link
  43. S. Hu, Y. Tu, X. Han, C. He, G. Cui, X. Long, Z. Zheng, Y. Fang, Y. Huang, W. Zhao et al., “MiniCPM: Unveiling the potential of small language models with scalable training strategies,” in Proc. Conference on Language Modeling (COLM), 2024. arXiv
  44. A. Hägele, E. Bakouch, A. Kosson, L. Ben Allal, L. Von Werra and M. Jaggi, “Scaling laws and compute-optimal training beyond fixed training durations,” in Advances in Neural Information Processing Systems 37, 2024. link
  45. M. Wortsman, G. Ilharco, J. W. Kim, M. Li, S. Kornblith, R. Roelofs, R. Gontijo Lopes, H. Hajishirzi, A. Farhadi, H. Namkoong and L. Schmidt, “Robust fine-tuning of zero-shot models,” in Proc. IEEE/CVF CVPR, 2022. doi
  46. M. Wortsman, G. Ilharco, S. Y. Gadre, R. Roelofs, R. Gontijo-Lopes, A. S. Morcos, H. Namkoong, A. Farhadi, Y. Carmon, S. Kornblith and L. Schmidt, “Model soups: Averaging weights of multiple fine-tuned models improves accuracy without increasing inference time,” in Proc. ICML, PMLR 162, 2022. link
  47. PyTorch contributors, “Automatic mixed precision examples,” PyTorch 2.15 documentation, 2026. link
  48. PyTorch contributors, “DistributedDataParallel,” PyTorch 2.15.0 source documentation, 2026. Official source.