[MegatronBERT P0] add PretrainedConfig and unit test

paddlenlp/transformers/__init__.py

@@ -110,6 +110,7 @@
 from .mbart.configuration import *
 from .megatronbert.modeling import *
 from .megatronbert.tokenizer import *
+from .megatronbert.configuration import *
 from .prophetnet.modeling import *
 from .prophetnet.tokenizer import *
 from .mobilebert.modeling import *

paddlenlp/transformers/megatronbert/configuration.py

@@ -0,0 +1,159 @@
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# 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.
+""" MBart model configuration"""
+from __future__ import annotations
+
+from paddlenlp.transformers.configuration_utils import PretrainedConfig
+
+__all__ = [
+    "MegatronBert_PRETRAINED_INIT_CONFIGURATION",
+    "MegatronBert_PRETRAINED_RESOURCE_FILES_MAP",
+    "MegatronBertConfig",
+]
+
+MegatronBert_PRETRAINED_INIT_CONFIGURATION = {
+    "megatronbert-cased": {
+        "attention_probs_dropout_prob": 0.1,
+        "hidden_act": "gelu",
+        "hidden_dropout_prob": 0.1,
+        "hidden_size": 1024,
+        "initializer_range": 0.02,
+        "intermediate_size": 4096,
+        "max_position_embeddings": 512,
+        "num_attention_heads": 16,
+        "num_hidden_layers": 24,
+        "type_vocab_size": 2,
+        "vocab_size": 29056,
+        "pad_token_id": 0,
+    },
+    "megatronbert-uncased": {
+        "attention_probs_dropout_prob": 0.1,
+        "hidden_act": "gelu",
+        "hidden_dropout_prob": 0.1,
+        "hidden_size": 1024,
+        "initializer_range": 0.02,
+        "intermediate_size": 4096,
+        "max_position_embeddings": 512,
+        "num_attention_heads": 16,
+        "num_hidden_layers": 24,
+        "type_vocab_size": 2,
+        "vocab_size": 30592,
+        "pad_token_id": 0,
+    },
+}
+
+MegatronBert_PRETRAINED_RESOURCE_FILES_MAP = {
+    "model_state": {
+        "megatronbert-cased": "http://bj.bcebos.com/paddlenlp/models/transformers/"
+        "megatron-bert/megatronbert-cased/model_state.pdparams",
+        "megatronbert-uncased": "http://bj.bcebos.com/paddlenlp/models/transformers/"
+        "megatron-bert/megatronbert-cased/model_state.pdparams",
+    }
+}
+
+
+class MegatronBertConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`MegatronBertModel`]. It is used to instantiate a
+    MEGATRON_BERT 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 MEGATRON_BERT
+    [nvidia/megatron-bert-uncased-345m](https://huggingface.co/nvidia/megatron-bert-uncased-345m) architecture.
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+    Args:
+        vocab_size (int):
+            Vocabulary size of `inputs_ids` in `MegatronBertModel`. Also is the vocab size of token embedding matrix.
+            Defines the number of different tokens that can be represented by the `inputs_ids` passed when calling `MegatronBert`.
+        hidden_size (int, optional):
+            Dimensionality of the encoder layer and pooler layer. Defaults to `1024`.
+        pad_token_id (int, optional):
+            The index of padding token in the token vocabulary.
+            Defaults to `0`.
+        type_vocab_size (int, optional):
+            The vocabulary size of `token_type_ids`.
+            Defaults to `2`.
+        hidden_act (str, optional):
+            The non-linear activation function in the feed-forward layer.
+            ``"gelu"``, ``"relu"`` and any other paddle supported activation functions
+            are supported. Defaults to `"gelu"`.
+        attention_probs_dropout_prob (float, optional):
+            The dropout probability used in MultiHeadAttention in all encoder layers to drop some attention target.
+            Defaults to `0.1`.
+        num_attention_heads (int, optional):
+            Number of attention heads for each attention layer in the Transformer encoder.
+            Defaults to `16`.
+        num_hidden_layers (int, optional):
+            Number of hidden layers in the Transformer encoder. Defaults to `24`.
+        max_position_embeddings (int, optional):
+            The maximum value of the dimensionality of position encoding, which dictates the maximum supported length of an input
+            sequence. Defaults to `512`.
+        hidden_dropout_prob (float, optional):
+            The dropout probability for all fully connected layers in the embeddings and encoder.
+            Defaults to `0.1`.
+        intermediate_size (int, optional):
+            Dimensionality of the feed-forward (ff) layer in the encoder. Input tensors
+            to ff layers are firstly projected from `hidden_size` to `intermediate_size`,
+            and then projected back to `hidden_size`. Typically `intermediate_size` is larger than `hidden_size`.
+            Defaults to `4096`.
+        position_embedding_type (str, optional):
+            Type of position embedding. Defaults to "absolute"
+        initializer_range (float, optional):
+            The standard deviation of the normal initializer.
+            Defaults to 0.02.
+
+            .. note::
+                A normal_initializer initializes weight matrices as normal distributions.
+                See :meth:`MegatronBertPretrainedModel.init_weights()` for how weights are initialized in `MegatronBertModel`.
+
+    """
+    model_type = "megatronbert"
+    keys_to_ignore_at_inference = ["past_key_values"]
+    attribute_map = {"num_attention_heads": "encoder_attention_heads", "hidden_size": "d_model"}
+
+    def __init__(
+        self,
+        vocab_size=29056,
+        hidden_size=1024,
+        num_hidden_layers=24,
+        num_attention_heads=16,
+        intermediate_size=4096,
+        hidden_act="gelu",
+        hidden_dropout_prob=0.1,
+        attention_probs_dropout_prob=0.1,
+        max_position_embeddings=512,
+        type_vocab_size=2,
+        initializer_range=0.02,
+        layer_norm_eps=1e-12,
+        pad_token_id=0,
+        position_embedding_type="absolute",
+        # use_cache=True,
+        **kwargs,
+    ):
+        super().__init__(pad_token_id=pad_token_id, **kwargs)
+
+        self.vocab_size = vocab_size
+        self.hidden_size = hidden_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.hidden_act = hidden_act
+        self.intermediate_size = intermediate_size
+        self.hidden_dropout_prob = hidden_dropout_prob
+        self.attention_probs_dropout_prob = attention_probs_dropout_prob
+        self.max_position_embeddings = max_position_embeddings
+        self.type_vocab_size = type_vocab_size
+        self.initializer_range = initializer_range
+        self.layer_norm_eps = layer_norm_eps
+        self.position_embedding_type = position_embedding_type
+        # self.use_cache = use_cache