# coding=utf-8
# Copyright 2020, The T5 Authors and HuggingFace Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" mT5 model configuration """
from ...configuration_utils import PretrainedConfig
from ...utils import logging
logger = logging.get_logger(__name__)
[docs]class MT5Config(PretrainedConfig):
r"""
This is the configuration class to store the configuration of a :class:`~transformers.MT5Model` or a
:class:`~transformers.TFMT5Model`. It is used to instantiate a mT5 model according to the specified arguments,
defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration
to that of the mT5 `google/mt5-small <https://huggingface.co/google/mt5-small>`__ architecture.
Configuration objects inherit from :class:`~transformers.PretrainedConfig` and can be used to control the model
outputs. Read the documentation from :class:`~transformers.PretrainedConfig` for more information.
Arguments:
vocab_size (:obj:`int`, `optional`, defaults to 32128):
Vocabulary size of the T5 model. Defines the number of different tokens that can be represented by the
:obj:`inputs_ids` passed when calling :class:`~transformers.T5Model` or :class:`~transformers.TFT5Model`.
d_model (:obj:`int`, `optional`, defaults to 512):
Size of the encoder layers and the pooler layer.
d_kv (:obj:`int`, `optional`, defaults to 64):
Size of the key, query, value projections per attention head. :obj:`d_kv` has to be equal to :obj:`d_model
// num_heads`.
d_ff (:obj:`int`, `optional`, defaults to 1024):
Size of the intermediate feed forward layer in each :obj:`T5Block`.
num_layers (:obj:`int`, `optional`, defaults to 8):
Number of hidden layers in the Transformer encoder.
num_decoder_layers (:obj:`int`, `optional`):
Number of hidden layers in the Transformer decoder. Will use the same value as :obj:`num_layers` if not
set.
num_heads (:obj:`int`, `optional`, defaults to 6):
Number of attention heads for each attention layer in the Transformer encoder.
relative_attention_num_buckets (:obj:`int`, `optional`, defaults to 32):
The number of buckets to use for each attention layer.
dropout_rate (:obj:`float`, `optional`, defaults to 0.1):
The ratio for all dropout layers.
layer_norm_eps (:obj:`float`, `optional`, defaults to 1e-6):
The epsilon used by the layer normalization layers.
initializer_factor (:obj:`float`, `optional`, defaults to 1):
A factor for initializing all weight matrices (should be kept to 1, used internally for initialization
testing).
feed_forward_proj (:obj:`string`, `optional`, defaults to :obj:`"gated-gelu"`):
Type of feed forward layer to be used. Should be one of :obj:`"relu"` or :obj:`"gated-gelu"`.
use_cache (:obj:`bool`, `optional`, defaults to :obj:`True`):
Whether or not the model should return the last key/values attentions (not used by all models).
"""
model_type = "mt5"
keys_to_ignore_at_inference = ["past_key_values"]
def __init__(
self,
vocab_size=250112,
d_model=512,
d_kv=64,
d_ff=1024,
num_layers=8,
num_decoder_layers=None,
num_heads=6,
relative_attention_num_buckets=32,
dropout_rate=0.1,
layer_norm_epsilon=1e-6,
initializer_factor=1.0,
feed_forward_proj="gated-gelu",
is_encoder_decoder=True,
use_cache=True,
tokenizer_class="T5Tokenizer",
tie_word_embeddings=False,
pad_token_id=0,
eos_token_id=1,
decoder_start_token_id=0,
**kwargs
):
super().__init__(
is_encoder_decoder=is_encoder_decoder,
tokenizer_class=tokenizer_class,
tie_word_embeddings=tie_word_embeddings,
pad_token_id=pad_token_id,
eos_token_id=eos_token_id,
decoder_start_token_id=decoder_start_token_id,
**kwargs,
)
self.vocab_size = vocab_size
self.d_model = d_model
self.d_kv = d_kv
self.d_ff = d_ff
self.num_layers = num_layers
self.num_decoder_layers = (
num_decoder_layers if num_decoder_layers is not None else self.num_layers
) # default = symmetry
self.num_heads = num_heads
self.relative_attention_num_buckets = relative_attention_num_buckets
self.dropout_rate = dropout_rate
self.layer_norm_epsilon = layer_norm_epsilon
self.initializer_factor = initializer_factor
self.feed_forward_proj = feed_forward_proj
self.use_cache = use_cache
@property
def hidden_size(self):
return self.d_model
@property
def num_attention_heads(self):
return self.num_heads
@property
def num_hidden_layers(self):
return self.num_layers