page_type: reference description: K-Means partitioning tree configuration.

<link rel="stylesheet" href="/site-assets/css/style.css"> <!-- DO NOT EDIT! Automatically generated file. --> <div itemscope itemtype="http://developers.google.com/ReferenceObject"> <meta itemprop="name" content="tflite_model_maker.searcher.Tree" /> <meta itemprop="path" content="Stable" /> <meta itemprop="property" content="__eq__"/> <meta itemprop="property" content="__init__"/> <meta itemprop="property" content="min_partition_size"/> <meta itemprop="property" content="quantize_centroids"/> <meta itemprop="property" content="random_init"/> <meta itemprop="property" content="spherical"/> <meta itemprop="property" content="training_iterations"/> <meta itemprop="property" content="training_sample_size"/> </div>

tflite_model_maker.searcher.Tree

<!-- Insert buttons and diff --> <table class="tfo-notebook-buttons tfo-api nocontent" align="left"> <td> <a target="_blank" href="https://github.com/tensorflow/examples/blob/master/tensorflow_examples/lite/model_maker/core/task/searcher.py#L45-L82"> </a> </td> </table>

K-Means partitioning tree configuration.

<pre class="devsite-click-to-copy prettyprint lang-py tfo-signature-link"> <code>tflite_model_maker.searcher.Tree( num_leaves: int, num_leaves_to_search: int, training_sample_size: int = 100000, min_partition_size: int = 50, training_iterations: int = 12, spherical: bool = False, quantize_centroids: bool = False, random_init: bool = True ) </code></pre> <h3>Used in the notebooks</h3> <table class="vertical-rules"> <thead> <tr> <th>Used in the tutorials</th> </tr> </thead> <tbody> <tr> <td> <ul> <li><a href="https://www.tensorflow.org/lite/models/modify/model_maker/text_searcher">Text Searcher with TensorFlow Lite Model Maker</a></li> </ul> </td> </tr> </tbody> </table>

In ScaNN, we use single layer K-Means tree to partition the database (index) as a way to reduce search space.

<!-- Tabular view --> <table class="responsive fixed orange"> <colgroup><col width="214px"><col></colgroup> <tr><th colspan="2"><h2 class="add-link">Attributes</h2></th></tr> <tr> <td> `num_leaves`<a id="num_leaves"></a> </td> <td> How many leaves (partitions) to have on the K-Means tree. In general, a good starting point would be the square root of the database size. </td> </tr><tr> <td> `num_leaves_to_search`<a id="num_leaves_to_search"></a> </td> <td> During inference ScaNN will compare the query vector against all the partition centroids and select the closest `num_leaves_to_search` ones to search in. The more leaves to search, the better the retrieval quality, and higher computational cost. </td> </tr><tr> <td> `training_sample_size`<a id="training_sample_size"></a> </td> <td> How many database embeddings to sample for the K-Means training. Generally, you want to use a large enough sample of the database to train K-Means so that it's representative enough. However, large sample can also lead to longer training time. A good starting value would be 100k, or the whole dataset if it's smaller than that. </td> </tr><tr> <td> `min_partition_size`<a id="min_partition_size"></a> </td> <td> Smallest allowable cluster size. Any clusters smaller than this will be removed, and its data points will be merged with other clusters. Recommended to be 1/10 of average cluster size (size of database divided by `num_leaves`) </td> </tr><tr> <td> `training_iterations`<a id="training_iterations"></a> </td> <td> How many itrations to train K-Means. </td> </tr><tr> <td> `spherical`<a id="spherical"></a> </td> <td> If true, L2 normalize the K-Means centroids. </td> </tr><tr> <td> `quantize_centroids`<a id="quantize_centroids"></a> </td> <td> If true, quantize centroids to int8. </td> </tr><tr> <td> `random_init`<a id="random_init"></a> </td> <td> If true, use random init. Otherwise use K-Means++. </td> </tr> </table>

Methods

<h3 id="__eq__"><code>__eq__</code></h3> <pre class="devsite-click-to-copy prettyprint lang-py tfo-signature-link"> <code>__eq__( other ) </code></pre> <!-- Tabular view --> <table class="responsive fixed orange"> <colgroup><col width="214px"><col></colgroup> <tr><th colspan="2"><h2 class="add-link">Class Variables</h2></th></tr> <tr> <td> min_partition_size<a id="min_partition_size"></a> </td> <td> `50` </td> </tr><tr> <td> quantize_centroids<a id="quantize_centroids"></a> </td> <td> `False` </td> </tr><tr> <td> random_init<a id="random_init"></a> </td> <td> `True` </td> </tr><tr> <td> spherical<a id="spherical"></a> </td> <td> `False` </td> </tr><tr> <td> training_iterations<a id="training_iterations"></a> </td> <td> `12` </td> </tr><tr> <td> training_sample_size<a id="training_sample_size"></a> </td> <td> `100000` </td> </tr> </table>