Step 5: Tune Hyperparameters
Stay organized with collections Save and categorize content based on your preferences.
We had to choose a number of hyperparameters for defining and training the model. We relied on intuition, examples and best practice recommendations. Our first choice of hyperparameter values, however, may not yield the best results. It only gives us a good starting point for training. Every problem is different and tuning these hyperparameters will help refine our model to better represent the particularities of the problem at hand. Let’s take a look at some of the hyperparameters we used and what it means to tune them:
Number of layers in the model: The number of layers in a neural network is an indicator of its complexity. We must be careful in choosing this value. Too many layers will allow the model to learn too much information about the training data, causing overfitting. Too few layers can limit the model’s learning ability, causing underfitting. For text classification datasets, we experimented with one, two, and three-layer MLPs. Models with two layers performed well, and in some cases better than three-layer models. Similarly, we tried sepCNNs with four and six layers, and the four-layer models performed well.
Number of units per layer: The units in a layer must hold the information for the transformation that a layer performs. For the first layer, this is driven by the number of features. In subsequent layers, the number of units depends on the choice of expanding or contracting the representation from the previous layer. Try to minimize the information loss between layers. We tried unit values in the range [8, 16, 32, 64], and 32/64 units worked well.
Dropout rate: Dropout layers are used in the model for regularization. They define the fraction of input to drop as a precaution for overfitting. Recommended range: 0.2–0.5.
Learning rate: This is the rate at which the neural network weights change between iterations. A large learning rate may cause large swings in the weights, and we may never find their optimal values. A low learning rate is good, but the model will take more iterations to converge. It is a good idea to start low, say at 1e-4. If the training is very slow, increase this value. If your model is not learning, try decreasing learning rate.
There are couple of additional hyperparameters we tuned that are specific to our sepCNN model:
Kernel size: The size of the convolution window. Recommended values: 3 or 5.
Embedding dimensions: The number of dimensions we want to use to represent word embeddings—i.e., the size of each word vector. Recommended values: 50–300. In our experiments, we used GloVe embeddings with 200 dimensions with a pre- trained embedding layer.
Play around with these hyperparameters and see what works best. Once you have chosen the best-performing hyperparameters for your use case, your model is ready to be deployed.
Except as otherwise noted, the content of this page is licensed under the Creative Commons Attribution 4.0 License, and code samples are licensed under the Apache 2.0 License. For details, see the Google Developers Site Policies. Java is a registered trademark of Oracle and/or its affiliates.
Last updated 2025-08-25 UTC.
[[["Easy to understand","easyToUnderstand","thumb-up"],["Solved my problem","solvedMyProblem","thumb-up"],["Other","otherUp","thumb-up"]],[["Missing the information I need","missingTheInformationINeed","thumb-down"],["Too complicated / too many steps","tooComplicatedTooManySteps","thumb-down"],["Out of date","outOfDate","thumb-down"],["Samples / code issue","samplesCodeIssue","thumb-down"],["Other","otherDown","thumb-down"]],["Last updated 2025-08-25 UTC."],[[["\u003cp\u003eInitial hyperparameter choices provide a starting point for model training, but further tuning is crucial to optimize performance for specific text classification problems.\u003c/p\u003e\n"],["\u003cp\u003eThe number of layers in a neural network impacts its complexity, with two-layer MLPs and four-layer sepCNNs showing promising results in text classification.\u003c/p\u003e\n"],["\u003cp\u003eKey hyperparameters to adjust include the number of units per layer (32 or 64 performed well), dropout rate (0.2-0.5 recommended), and learning rate (start low and adjust based on training progress).\u003c/p\u003e\n"],["\u003cp\u003eFor sepCNN models, optimizing kernel size (3 or 5) and embedding dimensions (50-300) further enhances performance.\u003c/p\u003e\n"],["\u003cp\u003eExperimenting with different hyperparameter combinations is essential to achieve the best model performance for your specific use case before deployment.\u003c/p\u003e\n"]]],[],null,["We had to choose a number of hyperparameters for defining and training the\nmodel. We relied on intuition, examples and best practice recommendations. Our\nfirst choice of hyperparameter values, however, may not yield the best results.\nIt only gives us a good starting point for training. Every problem is different\nand tuning these hyperparameters will help refine our model to better represent\nthe particularities of the problem at hand. Let's take a look at some of the\nhyperparameters we used and what it means to tune them:\n\n- **Number of layers in the model** : The number of layers in a neural network is\n an indicator of its complexity. We must be careful in choosing this value. Too\n many layers will allow the model to learn too much information about the\n training data, causing overfitting. Too few layers can limit the model's\n learning ability, causing underfitting. For text classification datasets, we\n experimented with one, two, and three-layer MLPs. Models with two layers\n performed well, and in some cases better than three-layer models. Similarly, we\n tried [sepCNN](https://developers.google.com/machine-learning/glossary?utm_source=DevSite&utm_campaign=Text-Class-Guide&utm_medium=referral&utm_content=glossary&utm_term=sepCNN#depthwise-separable-convolutional-neural-network-sepcnn)s\n with four and six layers, and the four-layer models performed well.\n\n- **Number of units per layer** : The units in a layer must hold the information\n for the transformation that a layer performs. For the first layer, this is\n driven by the number of features. In subsequent layers, the number of units\n depends on the choice of expanding or contracting the representation from the\n previous layer. Try to minimize the information loss between layers. We tried\n unit values in the range `[8, 16, 32, 64]`, and 32/64 units worked well.\n\n- **Dropout rate** : Dropout layers are used in the model for\n [regularization](https://developers.google.com/machine-learning/glossary/?utm_source=DevSite&utm_campaign=Text-Class-Guide&utm_medium=referral&utm_content=glossary&utm_term=dropout-regularization#dropout_regularization).\n They define the fraction of input to drop as a precaution for overfitting.\n Recommended range: 0.2--0.5.\n\n- **Learning rate**: This is the rate at which the neural network weights change\n between iterations. A large learning rate may cause large swings in the weights,\n and we may never find their optimal values. A low learning rate is good, but the\n model will take more iterations to converge. It is a good idea to start low, say\n at 1e-4. If the training is very slow, increase this value. If your model is not\n learning, try decreasing learning rate.\n\nThere are couple of additional hyperparameters we tuned that are specific to our\nsepCNN model:\n\n1. **Kernel size**: The size of the convolution window. Recommended values: 3 or\n 5.\n\n2. **Embedding dimensions**: The number of dimensions we want to use to represent\n word embeddings---i.e., the size of each word vector. Recommended values: 50--300.\n In our experiments, we used GloVe embeddings with 200 dimensions with a pre-\n trained embedding layer.\n\nPlay around with these hyperparameters and see what works best. Once you have\nchosen the best-performing hyperparameters for your use case, your model is\nready to be deployed."]]