ElectraBackbone
classkeras_nlp.models.ElectraBackbone(
vocab_size,
num_layers,
num_heads,
hidden_dim,
embedding_dim,
intermediate_dim,
dropout=0.1,
max_sequence_length=512,
num_segments=2,
dtype=None,
**kwargs
)
A Electra encoder network.
This network implements a bidirectional Transformer-based encoder as described in "Electra: Pre-training Text Encoders as Discriminators Rather Than Generators". It includes the embedding lookups and transformer layers, but not the masked language model or classification task networks.
The default constructor gives a fully customizable, randomly initialized
ELECTRA encoder with any number of layers, heads, and embedding
dimensions. To load preset architectures and weights, use the
from_preset()
constructor.
Disclaimer: Pre-trained models are provided on an "as is" basis, without warranties or conditions of any kind. The underlying model is provided by a third party and subject to a separate license, available here.
Arguments
max_sequence_length
uses the value from
sequence length. This determines the variable shape for positional
embeddings.keras.mixed_precision.DTypePolicy
. The dtype to use
for model computations and weights. Note that some computations,
such as softmax and layer normalization, will always be done at
float32 precision regardless of dtype.Example
input_data = {
"token_ids": np.ones(shape=(1, 12), dtype="int32"),
"segment_ids": np.array([[0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0]]),
"padding_mask": np.array([[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0]]),
}
# Pre-trained ELECTRA encoder.
model = keras_nlp.models.ElectraBackbone.from_preset(
"electra_base_discriminator_en"
)
model(input_data)
# Randomly initialized Electra encoder
backbone = keras_nlp.models.ElectraBackbone(
vocabulary_size=1000,
num_layers=2,
num_heads=2,
hidden_dim=32,
intermediate_dim=64,
dropout=0.1,
max_sequence_length=512,
)
# Returns sequence and pooled outputs.
sequence_output, pooled_output = backbone(input_data)
from_preset
methodElectraBackbone.from_preset(preset, load_weights=True, **kwargs)
Instantiate a keras_nlp.models.Backbone
from a model preset.
A preset is a directory of configs, weights and other file assets used
to save and load a pre-trained model. The preset
can be passed as a
one of:
'bert_base_en'
'kaggle://user/bert/keras/bert_base_en'
'hf://user/bert_base_en'
'./bert_base_en'
This constructor can be called in one of two ways. Either from the base
class like keras_nlp.models.Backbone.from_preset()
, or from
a model class like keras_nlp.models.GemmaBackbone.from_preset()
.
If calling from the base class, the subclass of the returning object
will be inferred from the config in the preset directory.
For any Backbone
subclass, you can run cls.presets.keys()
to list
all built-in presets available on the class.
Arguments
True
, the weights will be loaded into the
model architecture. If False
, the weights will be randomly
initialized.Examples
# Load a Gemma backbone with pre-trained weights.
model = keras_nlp.models.Backbone.from_preset(
"gemma_2b_en",
)
# Load a Bert backbone with a pre-trained config and random weights.
model = keras_nlp.models.Backbone.from_preset(
"bert_base_en",
load_weights=False,
)
Preset name | Parameters | Description |
---|---|---|
electra_small_discriminator_uncased_en | 13.55M | 12-layer small ELECTRA discriminator model. All inputs are lowercased. Trained on English Wikipedia + BooksCorpus. |
electra_small_generator_uncased_en | 13.55M | 12-layer small ELECTRA generator model. All inputs are lowercased. Trained on English Wikipedia + BooksCorpus. |
electra_base_discriminator_uncased_en | 109.48M | 12-layer base ELECTRA discriminator model. All inputs are lowercased. Trained on English Wikipedia + BooksCorpus. |
electra_base_generator_uncased_en | 33.58M | 12-layer base ELECTRA generator model. All inputs are lowercased. Trained on English Wikipedia + BooksCorpus. |
electra_large_discriminator_uncased_en | 335.14M | 24-layer large ELECTRA discriminator model. All inputs are lowercased. Trained on English Wikipedia + BooksCorpus. |
electra_large_generator_uncased_en | 51.07M | 24-layer large ELECTRA generator model. All inputs are lowercased. Trained on English Wikipedia + BooksCorpus. |
token_embedding
propertykeras_nlp.models.ElectraBackbone.token_embedding
A keras.layers.Embedding
instance for embedding token ids.
This layer embeds integer token ids to the hidden dim of the model.