Using text and neural network features

View on Run in Google Colab View on GitHub Download notebook See TF Hub model

Welcome to the Intermediate Colab for TensorFlow Decision Forests (TF-DF). In this colab, you will learn about some more advanced capabilities of TF-DF, including how to deal with natural language features.

This colab assumes you are familiar with the concepts presented the Beginner colab, notably about the installation about TF-DF.

In this colab, you will:

  1. Train a Random Forest that consumes text features natively as categorical sets.

  2. Train a Random Forest that consumes text features using a TensorFlow Hub module. In this setting (transfer learning), the module is already pre-trained on a large text corpus.

  3. Train a Gradient Boosted Decision Trees (GBDT) and a Neural Network together. The GBDT will consume the output of the Neural Network.


# Install TensorFlow Dececision Forests
pip install tensorflow_decision_forests

Wurlitzer is needed to display the detailed training logs in Colabs (when using verbose=2 in the model constructor).

pip install wurlitzer

Import the necessary libraries.

import tensorflow_decision_forests as tfdf

import os
import numpy as np
import pandas as pd
import tensorflow as tf
import math
2023-10-03 11:07:21.691323: E tensorflow/compiler/xla/stream_executor/cuda/] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
2023-10-03 11:07:21.691369: E tensorflow/compiler/xla/stream_executor/cuda/] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
2023-10-03 11:07:21.691401: E tensorflow/compiler/xla/stream_executor/cuda/] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered

The hidden code cell limits the output height in colab.

Use raw text as features

TF-DF can consume categorical-set features natively. Categorical-sets represent text features as bags of words (or n-grams).

For example: "The little blue dog"{"the", "little", "blue", "dog"}

In this example, you'll will train a Random Forest on the Stanford Sentiment Treebank (SST) dataset. The objective of this dataset is to classify sentences as carrying a positive or negative sentiment. You'll will use the binary classification version of the dataset curated in TensorFlow Datasets.

# Install the TensorFlow Datasets package
pip install tensorflow-datasets -U --quiet
# Load the dataset
import tensorflow_datasets as tfds
all_ds = tfds.load("glue/sst2")

# Display the first 3 examples of the test fold.
for example in all_ds["test"].take(3):
  print({attr_name: attr_tensor.numpy() for attr_name, attr_tensor in example.items()})
2023-10-03 11:07:27.781914: W tensorflow/core/common_runtime/gpu/] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at for how to download and setup the required libraries for your platform.
Skipping registering GPU devices...
{'idx': 163, 'label': -1, 'sentence': b'not even the hanson brothers can save it'}
{'idx': 131, 'label': -1, 'sentence': b'strong setup and ambitious goals fade as the film descends into unsophisticated scare tactics and b-film thuggery .'}
{'idx': 1579, 'label': -1, 'sentence': b'too timid to bring a sense of closure to an ugly chapter of the twentieth century .'}
2023-10-03 11:07:28.076020: W tensorflow/core/kernels/data/] The calling iterator did not fully read the dataset being cached. In order to avoid unexpected truncation of the dataset, the partially cached contents of the dataset  will be discarded. This can happen if you have an input pipeline similar to `dataset.cache().take(k).repeat()`. You should use `dataset.take(k).cache().repeat()` instead.

The dataset is modified as follows:

  1. The raw labels are integers in {-1, 1}, but the learning algorithm expects positive integer labels e.g. {0, 1}. Therefore, the labels are transformed as follows: new_labels = (original_labels + 1) / 2.
  2. A batch-size of 64 is applied to make reading the dataset more efficient.
  3. The sentence attribute needs to be tokenized, i.e. "hello world" -> ["hello", "world"].

Details: Some decision forest learning algorithms do not need a validation dataset (e.g. Random Forests) while others do (e.g. Gradient Boosted Trees in some cases). Since each learning algorithm under TF-DF can use validation data differently, TF-DF handles train/validation splits internally. As a result, when you have a training and validation sets, they can always be concatenated as input to the learning algorithm.

def prepare_dataset(example):
  label = (example["label"] + 1) // 2
  return {"sentence" : tf.strings.split(example["sentence"])}, label

train_ds = all_ds["train"].batch(100).map(prepare_dataset)
test_ds = all_ds["validation"].batch(100).map(prepare_dataset)

Finally, train and evaluate the model as usual. TF-DF automatically detects multi-valued categorical features as categorical-set.

%set_cell_height 300

# Specify the model.
model_1 = tfdf.keras.RandomForestModel(num_trees=30, verbose=2)

# Train the model.
<IPython.core.display.Javascript object>
Warning: The `num_threads` constructor argument is not set and the number of CPU is os.cpu_count()=32 > 32. Setting num_threads to 32. Set num_threads manually to use more than 32 cpus.
WARNING:absl:The `num_threads` constructor argument is not set and the number of CPU is os.cpu_count()=32 > 32. Setting num_threads to 32. Set num_threads manually to use more than 32 cpus.
Use /tmpfs/tmp/tmpq_ulv_ho as temporary training directory
Reading training dataset...
Training tensor examples:
Features: {'sentence': tf.RaggedTensor(values=Tensor("data:0", shape=(None,), dtype=string), row_splits=Tensor("data_1:0", shape=(None,), dtype=int64))}
Label: Tensor("data_2:0", shape=(None,), dtype=int64)
Weights: None
Normalized tensor features:
 {'sentence': SemanticTensor(semantic=<Semantic.CATEGORICAL_SET: 4>, tensor=tf.RaggedTensor(values=Tensor("data:0", shape=(None,), dtype=string), row_splits=Tensor("data_1:0", shape=(None,), dtype=int64)))}
Training dataset read in 0:00:04.363250. Found 67349 examples.
Training model...
Standard output detected as not visible to the user e.g. running in a notebook. Creating a training log redirection. If training gets stuck, try calling tfdf.keras.set_training_logs_redirection(False).
[INFO 23-10-03 11:07:32.5968 UTC] Start Yggdrasil model training
[INFO 23-10-03 11:07:32.5969 UTC] Collect training examples
[INFO 23-10-03 11:07:32.5969 UTC] Dataspec guide:
column_guides {
  column_name_pattern: "^__LABEL$"
  categorial {
    min_vocab_frequency: 0
    max_vocab_count: -1
default_column_guide {
  categorial {
    max_vocab_count: 2000
  discretized_numerical {
    maximum_num_bins: 255
ignore_columns_without_guides: false
detect_numerical_as_discretized_numerical: false

[INFO 23-10-03 11:07:32.5972 UTC] Number of batches: 674
[INFO 23-10-03 11:07:32.5973 UTC] Number of examples: 67349
[INFO 23-10-03 11:07:32.6372 UTC] 12816 item(s) have been pruned (i.e. they are considered out of dictionary) for the column sentence (2000 item(s) left) because min_value_count=5 and max_number_of_unique_values=2000
[INFO 23-10-03 11:07:32.6891 UTC] Training dataset:
Number of records: 67349
Number of columns: 2

Number of columns by type:
    CATEGORICAL_SET: 1 (50%)
    CATEGORICAL: 1 (50%)


    1: "sentence" CATEGORICAL_SET has-dict vocab-size:2001 num-oods:3595 (5.33787%) most-frequent:"the" 27205 (40.3941%)

    0: "__LABEL" CATEGORICAL integerized vocab-size:3 no-ood-item

    nas: Number of non-available (i.e. missing) values.
    ood: Out of dictionary.
    manually-defined: Attribute which type is manually defined by the user i.e. the type was not automatically inferred.
    tokenized: The attribute value is obtained through tokenization.
    has-dict: The attribute is attached to a string dictionary e.g. a categorical attribute stored as a string.
    vocab-size: Number of unique values.

[INFO 23-10-03 11:07:32.6892 UTC] Configure learner
[INFO 23-10-03 11:07:32.6894 UTC] Training config:
learner: "RANDOM_FOREST"
features: "^sentence$"
label: "^__LABEL$"
random_seed: 123456
metadata {
  framework: "TF Keras"
pure_serving_model: false
[yggdrasil_decision_forests.model.random_forest.proto.random_forest_config] {
  num_trees: 30
  decision_tree {
    max_depth: 16
    min_examples: 5
    in_split_min_examples_check: true
    keep_non_leaf_label_distribution: true
    num_candidate_attributes: 0
    missing_value_policy: GLOBAL_IMPUTATION
    allow_na_conditions: false
    categorical_set_greedy_forward {
      sampling: 0.1
      max_num_items: -1
      min_item_frequency: 1
    growing_strategy_local {
    categorical {
      cart {
    axis_aligned_split {
    internal {
      sorting_strategy: PRESORTED
    uplift {
      min_examples_in_treatment: 5
      split_score: KULLBACK_LEIBLER
  winner_take_all_inference: true
  compute_oob_performances: true
  compute_oob_variable_importances: false
  num_oob_variable_importances_permutations: 1
  bootstrap_training_dataset: true
  bootstrap_size_ratio: 1
  adapt_bootstrap_size_ratio_for_maximum_training_duration: false
  sampling_with_replacement: true

[INFO 23-10-03 11:07:32.6898 UTC] Deployment config:
cache_path: "/tmpfs/tmp/tmpq_ulv_ho/working_cache"
num_threads: 32
try_resume_training: true

[INFO 23-10-03 11:07:32.6901 UTC] Train model
[INFO 23-10-03 11:07:32.6907 UTC] Training random forest on 67349 example(s) and 1 feature(s).
[INFO 23-10-03 11:08:07.3587 UTC] Training of tree  1/30 (tree index:2) done accuracy:0.739076 logloss:9.40465
[INFO 23-10-03 11:08:17.4979 UTC] Training of tree  4/30 (tree index:27) done accuracy:0.752152 logloss:5.60982
[INFO 23-10-03 11:08:22.5343 UTC] Training of tree  14/30 (tree index:5) done accuracy:0.800167 logloss:2.23029
[INFO 23-10-03 11:08:23.8822 UTC] Training of tree  24/30 (tree index:1) done accuracy:0.816862 logloss:0.944891
[INFO 23-10-03 11:08:25.5138 UTC] Training of tree  30/30 (tree index:21) done accuracy:0.821274 logloss:0.854486
[INFO 23-10-03 11:08:25.5141 UTC] Final OOB metrics: accuracy:0.821274 logloss:0.854486
[INFO 23-10-03 11:08:25.5224 UTC] Export model in log directory: /tmpfs/tmp/tmpq_ulv_ho with prefix 80d0a32f092644d1
[INFO 23-10-03 11:08:25.5515 UTC] Save model in resources
[INFO 23-10-03 11:08:25.5541 UTC] Model self evaluation:
Number of predictions (without weights): 67349
Number of predictions (with weights): 67349
Label: __LABEL

Accuracy: 0.821274  CI95[W][0.818828 0.8237]
LogLoss: : 0.854486
ErrorRate: : 0.178726

Default Accuracy: : 0.557826
Default LogLoss: : 0.686445
Default ErrorRate: : 0.442174

Confusion Table:
   0      1      2
0  0      0      0
1  0  19593  10187
2  0   1850  35719
Total: 67349

One vs other classes:

[INFO 23-10-03 11:08:25.5779 UTC] Loading model from path /tmpfs/tmp/tmpq_ulv_ho/model/ with prefix 80d0a32f092644d1
[INFO 23-10-03 11:08:25.8369 UTC] Model loaded with 30 root(s), 43180 node(s), and 1 input feature(s).
[INFO 23-10-03 11:08:25.8369 UTC] Engine "RandomForestGeneric" built
[INFO 23-10-03 11:08:25.8369 UTC] Use fast generic engine
Model trained in 0:00:53.264767
Compiling model...
Model compiled.
<keras.src.callbacks.History at 0x7f514c2378b0>

In the previous logs, note that sentence is a CATEGORICAL_SET feature.

The model is evaluated as usual:

evaluation = model_1.evaluate(test_ds)

print(f"BinaryCrossentropyloss: {evaluation[0]}")
print(f"Accuracy: {evaluation[1]}")
9/9 [==============================] - 1s 4ms/step - loss: 0.0000e+00 - accuracy: 0.7638
BinaryCrossentropyloss: 0.0
Accuracy: 0.7637614607810974

The training logs looks are follow:

import matplotlib.pyplot as plt

logs = model_1.make_inspector().training_logs()
plt.plot([log.num_trees for log in logs], [log.evaluation.accuracy for log in logs])
plt.xlabel("Number of trees")
plt.ylabel("Out-of-bag accuracy")


More trees would probably be beneficial (I am sure of it because I tried :p).

Use a pretrained text embedding

The previous example trained a Random Forest using raw text features. This example will use a pre-trained TF-Hub embedding to convert text features into a dense embedding, and then train a Random Forest on top of it. In this situation, the Random Forest will only "see" the numerical output of the embedding (i.e. it will not see the raw text).

In this experiment, will use the Universal-Sentence-Encoder. Different pre-trained embeddings might be suited for different types of text (e.g. different language, different task) but also for other type of structured features (e.g. images).

The embedding module can be applied in one of two places:

  1. During the dataset preparation.
  2. In the pre-processing stage of the model.

The second option is often preferable: Packaging the embedding in the model makes the model easier to use (and harder to misuse).

First install TF-Hub:

pip install --upgrade tensorflow-hub

Unlike before, you don't need to tokenize the text.

def prepare_dataset(example):
  label = (example["label"] + 1) // 2
  return {"sentence" : example["sentence"]}, label

train_ds = all_ds["train"].batch(100).map(prepare_dataset)
test_ds = all_ds["validation"].batch(100).map(prepare_dataset)
%set_cell_height 300

import tensorflow_hub as hub
# NNLM ( is also a good choice.
hub_url = ""
embedding = hub.KerasLayer(hub_url)

sentence = tf.keras.layers.Input(shape=(), name="sentence", dtype=tf.string)
embedded_sentence = embedding(sentence)

raw_inputs = {"sentence": sentence}
processed_inputs = {"embedded_sentence": embedded_sentence}
preprocessor = tf.keras.Model(inputs=raw_inputs, outputs=processed_inputs)

model_2 = tfdf.keras.RandomForestModel(
<IPython.core.display.Javascript object>
Warning: The `num_threads` constructor argument is not set and the number of CPU is os.cpu_count()=32 > 32. Setting num_threads to 32. Set num_threads manually to use more than 32 cpus.
WARNING:absl:The `num_threads` constructor argument is not set and the number of CPU is os.cpu_count()=32 > 32. Setting num_threads to 32. Set num_threads manually to use more than 32 cpus.
Use /tmpfs/tmp/tmpsq__pjfx as temporary training directory
Reading training dataset...
Training dataset read in 0:00:10.889435. Found 67349 examples.
Training model...
[INFO 23-10-03 11:09:08.5640 UTC] Loading model from path /tmpfs/tmp/tmpsq__pjfx/model/ with prefix 3970687996794f0b
Model trained in 0:00:13.981656
Compiling model...
[INFO 23-10-03 11:09:10.3291 UTC] Model loaded with 100 root(s), 566000 node(s), and 512 input feature(s).
[INFO 23-10-03 11:09:10.3292 UTC] Engine "RandomForestOptPred" built
[INFO 23-10-03 11:09:10.3293 UTC] Use fast generic engine
Model compiled.
<keras.src.callbacks.History at 0x7f50598f50d0>
evaluation = model_2.evaluate(test_ds)

print(f"BinaryCrossentropyloss: {evaluation[0]}")
print(f"Accuracy: {evaluation[1]}")
9/9 [==============================] - 2s 16ms/step - loss: 0.0000e+00 - accuracy: 0.7913
BinaryCrossentropyloss: 0.0
Accuracy: 0.7912843823432922

Note that categorical sets represent text differently from a dense embedding, so it may be useful to use both strategies jointly.

Train a decision tree and neural network together

The previous example used a pre-trained Neural Network (NN) to process the text features before passing them to the Random Forest. This example will train both the Neural Network and the Random Forest from scratch.

TF-DF's Decision Forests do not back-propagate gradients (although this is the subject of ongoing research). Therefore, the training happens in two stages:

  1. Train the neural-network as a standard classification task:
example → [Normalize] → [Neural Network*] → [classification head] → prediction
*: Training.
  1. Replace the Neural Network's head (the last layer and the soft-max) with a Random Forest. Train the Random Forest as usual:
example → [Normalize] → [Neural Network] → [Random Forest*] → prediction
*: Training.

Prepare the dataset

This example uses the Palmer's Penguins dataset. See the Beginner colab for details.

First, download the raw data:

wget -q -O /tmp/penguins.csv

Load a dataset into a Pandas Dataframe.

dataset_df = pd.read_csv("/tmp/penguins.csv")

# Display the first 3 examples.

Prepare the dataset for training.

label = "species"

# Replaces numerical NaN (representing missing values in Pandas Dataframe) with 0s.
# ...Neural Nets don't work well with numerical NaNs.
for col in dataset_df.columns:
  if dataset_df[col].dtype not in [str, object]:
    dataset_df[col] = dataset_df[col].fillna(0)
# Split the dataset into a training and testing dataset.

def split_dataset(dataset, test_ratio=0.30):
  """Splits a panda dataframe in two."""
  test_indices = np.random.rand(len(dataset)) < test_ratio
  return dataset[~test_indices], dataset[test_indices]

train_ds_pd, test_ds_pd = split_dataset(dataset_df)
print("{} examples in training, {} examples for testing.".format(
    len(train_ds_pd), len(test_ds_pd)))

# Convert the datasets into tensorflow datasets
train_ds = tfdf.keras.pd_dataframe_to_tf_dataset(train_ds_pd, label=label)
test_ds = tfdf.keras.pd_dataframe_to_tf_dataset(test_ds_pd, label=label)
249 examples in training, 95 examples for testing.

Build the models

Next create the neural network model using Keras' functional style.

To keep the example simple this model only uses two inputs.

input_1 = tf.keras.Input(shape=(1,), name="bill_length_mm", dtype="float")
input_2 = tf.keras.Input(shape=(1,), name="island", dtype="string")

nn_raw_inputs = [input_1, input_2]

Use preprocessing layers to convert the raw inputs to inputs appropriate for the neural network.

# Normalization.
Normalization = tf.keras.layers.Normalization
CategoryEncoding = tf.keras.layers.CategoryEncoding
StringLookup = tf.keras.layers.StringLookup

values = train_ds_pd["bill_length_mm"].values[:, tf.newaxis]
input_1_normalizer = Normalization()

values = train_ds_pd["island"].values
input_2_indexer = StringLookup(max_tokens=32)

input_2_onehot = CategoryEncoding(output_mode="binary", max_tokens=32)

normalized_input_1 = input_1_normalizer(input_1)
normalized_input_2 = input_2_onehot(input_2_indexer(input_2))

nn_processed_inputs = [normalized_input_1, normalized_input_2]
WARNING:tensorflow:max_tokens is deprecated, please use num_tokens instead.
WARNING:tensorflow:max_tokens is deprecated, please use num_tokens instead.

Build the body of the neural network:

y = tf.keras.layers.Concatenate()(nn_processed_inputs)
y = tf.keras.layers.Dense(16, activation=tf.nn.relu6)(y)
last_layer = tf.keras.layers.Dense(8, activation=tf.nn.relu, name="last")(y)

# "3" for the three label classes. If it were a binary classification, the
# output dim would be 1.
classification_output = tf.keras.layers.Dense(3)(y)

nn_model = tf.keras.models.Model(nn_raw_inputs, classification_output)

This nn_model directly produces classification logits.

Next create a decision forest model. This will operate on the high level features that the neural network extracts in the last layer before that classification head.

# To reduce the risk of mistakes, group both the decision forest and the
# neural network in a single keras model.
nn_without_head = tf.keras.models.Model(inputs=nn_model.inputs, outputs=last_layer)
df_and_nn_model = tfdf.keras.RandomForestModel(preprocessing=nn_without_head)
Warning: The `num_threads` constructor argument is not set and the number of CPU is os.cpu_count()=32 > 32. Setting num_threads to 32. Set num_threads manually to use more than 32 cpus.
WARNING:absl:The `num_threads` constructor argument is not set and the number of CPU is os.cpu_count()=32 > 32. Setting num_threads to 32. Set num_threads manually to use more than 32 cpus.
Use /tmpfs/tmp/tmpmdyr1owp as temporary training directory

Train and evaluate the models

The model will be trained in two stages. First train the neural network with its own classification head:

%set_cell_height 300

  metrics=["accuracy"]), validation_data=test_ds, epochs=10)
<IPython.core.display.Javascript object>
Epoch 1/10
/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/keras/src/engine/ UserWarning: Input dict contained keys ['bill_depth_mm', 'flipper_length_mm', 'body_mass_g', 'sex', 'year'] which did not match any model input. They will be ignored by the model.
  inputs = self._flatten_to_reference_inputs(inputs)
1/1 [==============================] - 1s 622ms/step - loss: 1.0445 - accuracy: 0.3896 - val_loss: 1.0414 - val_accuracy: 0.4105
Epoch 2/10
1/1 [==============================] - 0s 18ms/step - loss: 1.0403 - accuracy: 0.3896 - val_loss: 1.0381 - val_accuracy: 0.4105
Epoch 3/10
1/1 [==============================] - 0s 17ms/step - loss: 1.0362 - accuracy: 0.3896 - val_loss: 1.0348 - val_accuracy: 0.4105
Epoch 4/10
1/1 [==============================] - 0s 16ms/step - loss: 1.0320 - accuracy: 0.3896 - val_loss: 1.0315 - val_accuracy: 0.4105
Epoch 5/10
1/1 [==============================] - 0s 17ms/step - loss: 1.0279 - accuracy: 0.3896 - val_loss: 1.0282 - val_accuracy: 0.4105
Epoch 6/10
1/1 [==============================] - 0s 16ms/step - loss: 1.0238 - accuracy: 0.3896 - val_loss: 1.0249 - val_accuracy: 0.4105
Epoch 7/10
1/1 [==============================] - 0s 16ms/step - loss: 1.0197 - accuracy: 0.3896 - val_loss: 1.0217 - val_accuracy: 0.4105
Epoch 8/10
1/1 [==============================] - 0s 16ms/step - loss: 1.0156 - accuracy: 0.3896 - val_loss: 1.0185 - val_accuracy: 0.4105
Epoch 9/10
1/1 [==============================] - 0s 16ms/step - loss: 1.0116 - accuracy: 0.3896 - val_loss: 1.0153 - val_accuracy: 0.4105
Epoch 10/10
1/1 [==============================] - 0s 16ms/step - loss: 1.0076 - accuracy: 0.3896 - val_loss: 1.0122 - val_accuracy: 0.4105
Model: "model_1"
 Layer (type)                Output Shape                 Param #   Connected to                  
 island (InputLayer)         [(None, 1)]                  0         []                            
 bill_length_mm (InputLayer  [(None, 1)]                  0         []                            
 string_lookup (StringLooku  (None, 1)                    0         ['island[0][0]']              
 normalization (Normalizati  (None, 1)                    3         ['bill_length_mm[0][0]']      
 category_encoding (Categor  (None, 32)                   0         ['string_lookup[0][0]']       
 concatenate (Concatenate)   (None, 33)                   0         ['normalization[0][0]',       
 dense (Dense)               (None, 16)                   544       ['concatenate[0][0]']         
 dense_1 (Dense)             (None, 3)                    51        ['dense[0][0]']               
Total params: 598 (2.34 KB)
Trainable params: 595 (2.32 KB)
Non-trainable params: 3 (16.00 Byte)

The neural network layers are shared between the two models. So now that the neural network is trained the decision forest model will be fit to the trained output of the neural network layers:

%set_cell_height 300
<IPython.core.display.Javascript object>
Reading training dataset...
Training dataset read in 0:00:00.249128. Found 249 examples.
Training model...
Model trained in 0:00:00.040728
Compiling model...
Model compiled.
[INFO 23-10-03 11:09:17.0249 UTC] Loading model from path /tmpfs/tmp/tmpmdyr1owp/model/ with prefix 4c794bd9ef944db7
[INFO 23-10-03 11:09:17.0390 UTC] Model loaded with 300 root(s), 4766 node(s), and 7 input feature(s).
[INFO 23-10-03 11:09:17.0391 UTC] Use fast generic engine
<keras.src.callbacks.History at 0x7f5059912610>

Now evaluate the composed model:

print("Evaluation:", df_and_nn_model.evaluate(test_ds))
1/1 [==============================] - 0s 141ms/step - loss: 0.0000e+00 - accuracy: 0.9474
Evaluation: [0.0, 0.9473684430122375]

Compare it to the Neural Network alone:

print("Evaluation :", nn_model.evaluate(test_ds))
1/1 [==============================] - 0s 10ms/step - loss: 1.0122 - accuracy: 0.4105
Evaluation : [1.0121650695800781, 0.410526305437088]