# coding=utf-8
# Copyright 2018 XXX Authors
#
# 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.
""" PyTorch XXX model. """

####################################################
# In this template, replace all the XXX (various casings) with your model name
####################################################

from __future__ import absolute_import, division, print_function, unicode_literals

import copy
import itertools
import json
import logging
import math
import os
import sys
from io import open

import torch
from torch import nn
from torch.nn import CrossEntropyLoss, MSELoss

from .configuration_xxx import XxxConfig
from .file_utils import add_start_docstrings
from .modeling_utils import PreTrainedModel, prune_linear_layer


logger = logging.getLogger(__name__)

####################################################
# This dict contrains shortcut names and associated url
# for the pretrained weights provided with the models
####################################################
XXX_PRETRAINED_MODEL_ARCHIVE_MAP = {
    "xxx-base-uncased": "https://s3.amazonaws.com/models.huggingface.co/bert/xxx-base-uncased-pytorch_model.bin",
    "xxx-large-uncased": "https://s3.amazonaws.com/models.huggingface.co/bert/xxx-large-uncased-pytorch_model.bin",
}


####################################################
# This is a conversion method from TF 1.0 to PyTorch
# More details: https://medium.com/huggingface/from-tensorflow-to-pytorch-265f40ef2a28
####################################################
def load_tf_weights_in_xxx(model, config, tf_checkpoint_path):
    """ Load tf checkpoints in a pytorch model.
    """
    try:
        import re
        import numpy as np
        import tensorflow as tf
    except ImportError:
        logger.error(
            "Loading a TensorFlow model in PyTorch, requires TensorFlow to be installed. Please see "
            "https://www.tensorflow.org/install/ for installation instructions."
        )
        raise
    tf_path = os.path.abspath(tf_checkpoint_path)
    logger.info("Converting TensorFlow checkpoint from {}".format(tf_path))
    # Load weights from TF model
    init_vars = tf.train.list_variables(tf_path)
    names = []
    arrays = []
    for name, shape in init_vars:
        logger.info("Loading TF weight {} with shape {}".format(name, shape))
        array = tf.train.load_variable(tf_path, name)
        names.append(name)
        arrays.append(array)

    for name, array in zip(names, arrays):
        name = name.split("/")
        # adam_v and adam_m are variables used in AdamWeightDecayOptimizer to calculated m and v
        # which are not required for using pretrained model
        if any(n in ["adam_v", "adam_m", "global_step"] for n in name):
            logger.info("Skipping {}".format("/".join(name)))
            continue
        pointer = model
        for m_name in name:
            if re.fullmatch(r"[A-Za-z]+_\d+", m_name):
                l = re.split(r"_(\d+)", m_name)
            else:
                l = [m_name]
            if l[0] == "kernel" or l[0] == "gamma":
                pointer = getattr(pointer, "weight")
            elif l[0] == "output_bias" or l[0] == "beta":
                pointer = getattr(pointer, "bias")
            elif l[0] == "output_weights":
                pointer = getattr(pointer, "weight")
            elif l[0] == "squad":
                pointer = getattr(pointer, "classifier")
            else:
                try:
                    pointer = getattr(pointer, l[0])
                except AttributeError:
                    logger.info("Skipping {}".format("/".join(name)))
                    continue
            if len(l) >= 2:
                num = int(l[1])
                pointer = pointer[num]
        if m_name[-11:] == "_embeddings":
            pointer = getattr(pointer, "weight")
        elif m_name == "kernel":
            array = np.transpose(array)
        try:
            assert pointer.shape == array.shape
        except AssertionError as e:
            e.args += (pointer.shape, array.shape)
            raise
        logger.info("Initialize PyTorch weight {}".format(name))
        pointer.data = torch.from_numpy(array)
    return model


####################################################
# PyTorch Models are constructed by sub-classing
# - torch.nn.Module for the layers and
# - PreTrainedModel for the models (itself a sub-class of torch.nn.Module)
####################################################

####################################################
# Here is an example of typical layer in a PyTorch model of the library
# The classes are usually identical to the TF 2.0 ones without the 'TF' prefix.
#
# See the conversion methods in modeling_tf_pytorch_utils.py for more details
####################################################
class XxxLayer(nn.Module):
    def __init__(self, config):
        super(XxxLayer, self).__init__()
        self.attention = XxxAttention(config)
        self.intermediate = XxxIntermediate(config)
        self.output = XxxOutput(config)

    def forward(self, hidden_states, attention_mask=None, head_mask=None):
        attention_outputs = self.attention(hidden_states, attention_mask, head_mask)
        attention_output = attention_outputs[0]
        intermediate_output = self.intermediate(attention_output)
        layer_output = self.output(intermediate_output, attention_output)
        outputs = (layer_output,) + attention_outputs[1:]  # add attentions if we output them
        return outputs


####################################################
# PreTrainedModel is a sub-class of torch.nn.Module
# which take care of loading and saving pretrained weights
# and various common utilities.
#
# Here you just need to specify a few (self-explanatory)
# pointers for your model and the weights initialization
# method if its not fully covered by PreTrainedModel's default method
####################################################
class XxxPreTrainedModel(PreTrainedModel):
    """ An abstract class to handle weights initialization and
        a simple interface for dowloading and loading pretrained models.
    """

    config_class = XxxConfig
    pretrained_model_archive_map = XXX_PRETRAINED_MODEL_ARCHIVE_MAP
    load_tf_weights = load_tf_weights_in_xxx
    base_model_prefix = "transformer"

    def _init_weights(self, module):
        """ Initialize the weights """
        if isinstance(module, (nn.Linear, nn.Embedding)):
            # Slightly different from the TF version which uses truncated_normal for initialization
            # cf https://github.com/pytorch/pytorch/pull/5617
            module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
        elif isinstance(module, XxxLayerNorm):
            module.bias.data.zero_()
            module.weight.data.fill_(1.0)
        if isinstance(module, nn.Linear) and module.bias is not None:
            module.bias.data.zero_()


XXX_START_DOCSTRING = r"""    The XXX model was proposed in
    `XXX: Pre-training of Deep Bidirectional Transformers for Language Understanding`_
    by Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova. It's a bidirectional transformer
    pre-trained using a combination of masked language modeling objective and next sentence prediction
    on a large corpus comprising the Toronto Book Corpus and Wikipedia.

    This model is a PyTorch `torch.nn.Module`_ sub-class. Use it as a regular PyTorch Module and
    refer to the PyTorch documentation for all matter related to general usage and behavior.

    .. _`XXX: Pre-training of Deep Bidirectional Transformers for Language Understanding`:
        https://arxiv.org/abs/1810.04805

    .. _`torch.nn.Module`:
        https://pytorch.org/docs/stable/nn.html#module

    Parameters:
        config (:class:`~transformers.XxxConfig`): Model configuration class with all the parameters of the model.
            Initializing with a config file does not load the weights associated with the model, only the configuration.
            Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model weights.
"""

XXX_INPUTS_DOCSTRING = r"""
    Inputs:
        **input_ids**: ``torch.LongTensor`` of shape ``(batch_size, sequence_length)``:
            Indices of input sequence tokens in the vocabulary.
            To match pre-training, XXX input sequence should be formatted with [CLS] and [SEP] tokens as follows:

            (a) For sequence pairs:

                ``tokens:         [CLS] is this jack ##son ##ville ? [SEP] no it is not . [SEP]``

                ``token_type_ids:   0   0  0    0    0     0       0   0   1  1  1  1   1   1``

            (b) For single sequences:

                ``tokens:         [CLS] the dog is hairy . [SEP]``

                ``token_type_ids:   0   0   0   0  0     0   0``

            Xxx is a model with absolute position embeddings so it's usually advised to pad the inputs on
            the right rather than the left.

            Indices can be obtained using :class:`transformers.XxxTokenizer`.
            See :func:`transformers.PreTrainedTokenizer.encode` and
            :func:`transformers.PreTrainedTokenizer.convert_tokens_to_ids` for details.
        **attention_mask**: (`optional`) ``torch.FloatTensor`` of shape ``(batch_size, sequence_length)``:
            Mask to avoid performing attention on padding token indices.
            Mask values selected in ``[0, 1]``:
            ``1`` for tokens that are NOT MASKED, ``0`` for MASKED tokens.
        **token_type_ids**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size, sequence_length)``:
            Segment token indices to indicate first and second portions of the inputs.
            Indices are selected in ``[0, 1]``: ``0`` corresponds to a `sentence A` token, ``1``
            corresponds to a `sentence B` token
            (see `XXX: Pre-training of Deep Bidirectional Transformers for Language Understanding`_ for more details).
        **position_ids**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size, sequence_length)``:
            Indices of positions of each input sequence tokens in the position embeddings.
            Selected in the range ``[0, config.max_position_embeddings - 1]``.
        **head_mask**: (`optional`) ``torch.FloatTensor`` of shape ``(num_heads,)`` or ``(num_layers, num_heads)``:
            Mask to nullify selected heads of the self-attention modules.
            Mask values selected in ``[0, 1]``:
            ``1`` indicates the head is **not masked**, ``0`` indicates the head is **masked**.
        **inputs_embeds**: (`optional`) ``torch.FloatTensor`` of shape ``(batch_size, sequence_length, embedding_dim)``:
            Optionally, instead of passing ``input_ids`` you can choose to directly pass an embedded representation.
            This is useful if you want more control over how to convert `input_ids` indices into associated vectors
            than the model's internal embedding lookup matrix.
"""


@add_start_docstrings(
    "The bare Xxx Model transformer outputting raw hidden-states without any specific head on top.",
    XXX_START_DOCSTRING,
    XXX_INPUTS_DOCSTRING,
)
class XxxModel(XxxPreTrainedModel):
    r"""
    Outputs: `Tuple` comprising various elements depending on the configuration (config) and inputs:
        **last_hidden_state**: ``torch.FloatTensor`` of shape ``(batch_size, sequence_length, hidden_size)``
            Sequence of hidden-states at the output of the last layer of the model.
        **pooler_output**: ``torch.FloatTensor`` of shape ``(batch_size, hidden_size)``
            Last layer hidden-state of the first token of the sequence (classification token)
            further processed by a Linear layer and a Tanh activation function. The Linear
            layer weights are trained from the next sentence prediction (classification)
            objective during Xxx pretraining. This output is usually *not* a good summary
            of the semantic content of the input, you're often better with averaging or pooling
            the sequence of hidden-states for the whole input sequence.
        **hidden_states**: (`optional`, returned when ``config.output_hidden_states=True``)
            list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
            of shape ``(batch_size, sequence_length, hidden_size)``:
            Hidden-states of the model at the output of each layer plus the initial embedding outputs.
        **attentions**: (`optional`, returned when ``config.output_attentions=True``)
            list of ``torch.FloatTensor`` (one for each layer) of shape ``(batch_size, num_heads, sequence_length, sequence_length)``:
            Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

    Examples::

        tokenizer = XxxTokenizer.from_pretrained('xxx-base-uncased')
        model = XxxModel.from_pretrained('xxx-base-uncased')
        input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute")).unsqueeze(0)  # Batch size 1
        outputs = model(input_ids)
        last_hidden_states = outputs[0]  # The last hidden-state is the first element of the output tuple

    """

    def __init__(self, config):
        super(XxxModel, self).__init__(config)

        self.embeddings = XxxEmbeddings(config)
        self.encoder = XxxEncoder(config)
        self.pooler = XxxPooler(config)

        self.init_weights()

    def get_input_embeddings(self):
        return self.embeddings.word_embeddings

    def set_input_embeddings(self, new_embeddings):
        self.embeddings.word_embeddings = new_embeddings

    def _prune_heads(self, heads_to_prune):
        """ Prunes heads of the model.
            heads_to_prune: dict of {layer_num: list of heads to prune in this layer}
            See base class PreTrainedModel
        """
        for layer, heads in heads_to_prune.items():
            self.encoder.layer[layer].attention.prune_heads(heads)

    def forward(
        self,
        input_ids=None,
        attention_mask=None,
        token_type_ids=None,
        position_ids=None,
        head_mask=None,
        inputs_embeds=None,
    ):
        if input_ids is not None and inputs_embeds is not None:
            raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
        elif input_ids is not None:
            input_shape = input_ids.size()
        elif inputs_embeds is not None:
            input_shape = inputs_embeds.size()[:-1]
        else:
            raise ValueError("You have to specify either input_ids or inputs_embeds")

        device = input_ids.device if input_ids is not None else inputs_embeds.device

        if attention_mask is None:
            attention_mask = torch.ones(input_shape, device=device)
        if token_type_ids is None:
            token_type_ids = torch.zeros(input_shape, dtype=torch.long, device=device)

        # We create a 3D attention mask from a 2D tensor mask.
        # Sizes are [batch_size, 1, 1, to_seq_length]
        # So we can broadcast to [batch_size, num_heads, from_seq_length, to_seq_length]
        # this attention mask is more simple than the triangular masking of causal attention
        # used in OpenAI GPT, we just need to prepare the broadcast dimension here.
        extended_attention_mask = attention_mask.unsqueeze(1).unsqueeze(2)

        # Since attention_mask is 1.0 for positions we want to attend and 0.0 for
        # masked positions, this operation will create a tensor which is 0.0 for
        # positions we want to attend and -10000.0 for masked positions.
        # Since we are adding it to the raw scores before the softmax, this is
        # effectively the same as removing these entirely.
        extended_attention_mask = extended_attention_mask.to(dtype=next(self.parameters()).dtype)  # fp16 compatibility
        extended_attention_mask = (1.0 - extended_attention_mask) * -10000.0

        # Prepare head mask if needed
        # 1.0 in head_mask indicate we keep the head
        # attention_probs has shape bsz x n_heads x N x N
        # input head_mask has shape [num_heads] or [num_hidden_layers x num_heads]
        # and head_mask is converted to shape [num_hidden_layers x batch x num_heads x seq_length x seq_length]
        if head_mask is not None:
            if head_mask.dim() == 1:
                head_mask = head_mask.unsqueeze(0).unsqueeze(0).unsqueeze(-1).unsqueeze(-1)
                head_mask = head_mask.expand(self.config.num_hidden_layers, -1, -1, -1, -1)
            elif head_mask.dim() == 2:
                head_mask = (
                    head_mask.unsqueeze(1).unsqueeze(-1).unsqueeze(-1)
                )  # We can specify head_mask for each layer
            head_mask = head_mask.to(
                dtype=next(self.parameters()).dtype
            )  # switch to fload if need + fp16 compatibility
        else:
            head_mask = [None] * self.config.num_hidden_layers

        ##################################
        # Replace this with your model code
        embedding_output = self.embeddings(
            input_ids=input_ids, position_ids=position_ids, token_type_ids=token_type_ids, inputs_embeds=inputs_embeds
        )
        encoder_outputs = self.encoder(embedding_output, extended_attention_mask, head_mask=head_mask)
        sequence_output = encoder_outputs[0]
        outputs = (sequence_output,) + encoder_outputs[1:]  # add hidden_states and attentions if they are here

        return outputs  # sequence_output, (hidden_states), (attentions)


@add_start_docstrings(
    """Xxx Model with a `language modeling` head on top. """, XXX_START_DOCSTRING, XXX_INPUTS_DOCSTRING
)
class XxxForMaskedLM(XxxPreTrainedModel):
    r"""
        **masked_lm_labels**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size, sequence_length)``:
            Labels for computing the masked language modeling loss.
            Indices should be in ``[-1, 0, ..., config.vocab_size]`` (see ``input_ids`` docstring)
            Tokens with indices set to ``-100`` are ignored (masked), the loss is only computed for the tokens with labels
            in ``[0, ..., config.vocab_size]``

    Outputs: `Tuple` comprising various elements depending on the configuration (config) and inputs:
        **loss**: (`optional`, returned when ``masked_lm_labels`` is provided) ``torch.FloatTensor`` of shape ``(1,)``:
            Masked language modeling loss.
        **prediction_scores**: ``torch.FloatTensor`` of shape ``(batch_size, sequence_length, config.vocab_size)``
            Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
        **hidden_states**: (`optional`, returned when ``config.output_hidden_states=True``)
            list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
            of shape ``(batch_size, sequence_length, hidden_size)``:
            Hidden-states of the model at the output of each layer plus the initial embedding outputs.
        **attentions**: (`optional`, returned when ``config.output_attentions=True``)
            list of ``torch.FloatTensor`` (one for each layer) of shape ``(batch_size, num_heads, sequence_length, sequence_length)``:
            Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

    Examples::

        tokenizer = XxxTokenizer.from_pretrained('xxx-base-uncased')
        model = XxxForMaskedLM.from_pretrained('xxx-base-uncased')
        input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute")).unsqueeze(0)  # Batch size 1
        outputs = model(input_ids, masked_lm_labels=input_ids)
        loss, prediction_scores = outputs[:2]

    """

    def __init__(self, config):
        super(XxxForMaskedLM, self).__init__(config)

        self.transformer = XxxModel(config)
        self.lm_head = nn.Linear(config.n_embd, config.vocab_size)

        self.init_weights()

    def get_output_embeddings(self):
        return self.lm_head

    def forward(
        self,
        input_ids=None,
        attention_mask=None,
        token_type_ids=None,
        position_ids=None,
        head_mask=None,
        inputs_embeds=None,
        masked_lm_labels=None,
    ):

        outputs = self.transformer(
            input_ids,
            attention_mask=attention_mask,
            token_type_ids=token_type_ids,
            position_ids=position_ids,
            head_mask=head_mask,
            inputs_embeds=inputs_embeds,
        )

        sequence_output = outputs[0]
        prediction_scores = self.cls(sequence_output)

        outputs = (prediction_scores,) + outputs[2:]  # Add hidden states and attention if they are here
        if masked_lm_labels is not None:
            loss_fct = CrossEntropyLoss()
            masked_lm_loss = loss_fct(prediction_scores.view(-1, self.config.vocab_size), masked_lm_labels.view(-1))
            outputs = (masked_lm_loss,) + outputs

        return outputs  # (masked_lm_loss), prediction_scores, (hidden_states), (attentions)


@add_start_docstrings(
    """Xxx Model transformer with a sequence classification/regression head on top (a linear layer on top of
    the pooled output) e.g. for GLUE tasks. """,
    XXX_START_DOCSTRING,
    XXX_INPUTS_DOCSTRING,
)
class XxxForSequenceClassification(XxxPreTrainedModel):
    r"""
        **labels**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size,)``:
            Labels for computing the sequence classification/regression loss.
            Indices should be in ``[0, ..., config.num_labels - 1]``.
            If ``config.num_labels == 1`` a regression loss is computed (Mean-Square loss),
            If ``config.num_labels > 1`` a classification loss is computed (Cross-Entropy).

    Outputs: `Tuple` comprising various elements depending on the configuration (config) and inputs:
        **loss**: (`optional`, returned when ``labels`` is provided) ``torch.FloatTensor`` of shape ``(1,)``:
            Classification (or regression if config.num_labels==1) loss.
        **logits**: ``torch.FloatTensor`` of shape ``(batch_size, config.num_labels)``
            Classification (or regression if config.num_labels==1) scores (before SoftMax).
        **hidden_states**: (`optional`, returned when ``config.output_hidden_states=True``)
            list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
            of shape ``(batch_size, sequence_length, hidden_size)``:
            Hidden-states of the model at the output of each layer plus the initial embedding outputs.
        **attentions**: (`optional`, returned when ``config.output_attentions=True``)
            list of ``torch.FloatTensor`` (one for each layer) of shape ``(batch_size, num_heads, sequence_length, sequence_length)``:
            Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

    Examples::

        tokenizer = XxxTokenizer.from_pretrained('xxx-base-uncased')
        model = XxxForSequenceClassification.from_pretrained('xxx-base-uncased')
        input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute")).unsqueeze(0)  # Batch size 1
        labels = torch.tensor([1]).unsqueeze(0)  # Batch size 1
        outputs = model(input_ids, labels=labels)
        loss, logits = outputs[:2]

    """

    def __init__(self, config):
        super(XxxForSequenceClassification, self).__init__(config)
        self.num_labels = config.num_labels

        self.transformer = XxxModel(config)
        self.dropout = nn.Dropout(config.hidden_dropout_prob)
        self.classifier = nn.Linear(config.hidden_size, self.config.num_labels)

        self.init_weights()

    def forward(
        self,
        input_ids=None,
        attention_mask=None,
        token_type_ids=None,
        position_ids=None,
        head_mask=None,
        inputs_embeds=None,
        labels=None,
    ):

        outputs = self.transformer(
            input_ids,
            attention_mask=attention_mask,
            token_type_ids=token_type_ids,
            position_ids=position_ids,
            head_mask=head_mask,
            inputs_embeds=inputs_embeds,
        )

        pooled_output = outputs[1]

        pooled_output = self.dropout(pooled_output)
        logits = self.classifier(pooled_output)

        outputs = (logits,) + outputs[2:]  # add hidden states and attention if they are here

        if labels is not None:
            if self.num_labels == 1:
                #  We are doing regression
                loss_fct = MSELoss()
                loss = loss_fct(logits.view(-1), labels.view(-1))
            else:
                loss_fct = CrossEntropyLoss()
                loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1))
            outputs = (loss,) + outputs

        return outputs  # (loss), logits, (hidden_states), (attentions)


@add_start_docstrings(
    """Xxx Model with a token classification head on top (a linear layer on top of
    the hidden-states output) e.g. for Named-Entity-Recognition (NER) tasks. """,
    XXX_START_DOCSTRING,
    XXX_INPUTS_DOCSTRING,
)
class XxxForTokenClassification(XxxPreTrainedModel):
    r"""
        **labels**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size, sequence_length)``:
            Labels for computing the token classification loss.
            Indices should be in ``[0, ..., config.num_labels - 1]``.

    Outputs: `Tuple` comprising various elements depending on the configuration (config) and inputs:
        **loss**: (`optional`, returned when ``labels`` is provided) ``torch.FloatTensor`` of shape ``(1,)``:
            Classification loss.
        **scores**: ``torch.FloatTensor`` of shape ``(batch_size, sequence_length, config.num_labels)``
            Classification scores (before SoftMax).
        **hidden_states**: (`optional`, returned when ``config.output_hidden_states=True``)
            list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
            of shape ``(batch_size, sequence_length, hidden_size)``:
            Hidden-states of the model at the output of each layer plus the initial embedding outputs.
        **attentions**: (`optional`, returned when ``config.output_attentions=True``)
            list of ``torch.FloatTensor`` (one for each layer) of shape ``(batch_size, num_heads, sequence_length, sequence_length)``:
            Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

    Examples::

        tokenizer = XxxTokenizer.from_pretrained('xxx-base-uncased')
        model = XxxForTokenClassification.from_pretrained('xxx-base-uncased')
        input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute")).unsqueeze(0)  # Batch size 1
        labels = torch.tensor([1] * input_ids.size(1)).unsqueeze(0)  # Batch size 1
        outputs = model(input_ids, labels=labels)
        loss, scores = outputs[:2]

    """

    def __init__(self, config):
        super(XxxForTokenClassification, self).__init__(config)
        self.num_labels = config.num_labels

        self.transformer = XxxModel(config)
        self.dropout = nn.Dropout(config.hidden_dropout_prob)
        self.classifier = nn.Linear(config.hidden_size, config.num_labels)

        self.init_weights()

    def forward(
        self,
        input_ids=None,
        attention_mask=None,
        token_type_ids=None,
        position_ids=None,
        head_mask=None,
        inputs_embeds=None,
        labels=None,
    ):

        outputs = self.transformer(
            input_ids,
            attention_mask=attention_mask,
            token_type_ids=token_type_ids,
            position_ids=position_ids,
            head_mask=head_mask,
            inputs_embeds=inputs_embeds,
        )

        sequence_output = outputs[0]

        sequence_output = self.dropout(sequence_output)
        logits = self.classifier(sequence_output)

        outputs = (logits,) + outputs[2:]  # add hidden states and attention if they are here
        if labels is not None:
            loss_fct = CrossEntropyLoss()
            # Only keep active parts of the loss
            if attention_mask is not None:
                active_loss = attention_mask.view(-1) == 1
                active_logits = logits.view(-1, self.num_labels)[active_loss]
                active_labels = labels.view(-1)[active_loss]
                loss = loss_fct(active_logits, active_labels)
            else:
                loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1))
            outputs = (loss,) + outputs

        return outputs  # (loss), scores, (hidden_states), (attentions)


@add_start_docstrings(
    """Xxx Model with a span classification head on top for extractive question-answering tasks like SQuAD (a linear layers on top of
    the hidden-states output to compute `span start logits` and `span end logits`). """,
    XXX_START_DOCSTRING,
    XXX_INPUTS_DOCSTRING,
)
class XxxForQuestionAnswering(XxxPreTrainedModel):
    r"""
        **start_positions**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size,)``:
            Labels for position (index) of the start of the labelled span for computing the token classification loss.
            Positions are clamped to the length of the sequence (`sequence_length`).
            Position outside of the sequence are not taken into account for computing the loss.
        **end_positions**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size,)``:
            Labels for position (index) of the end of the labelled span for computing the token classification loss.
            Positions are clamped to the length of the sequence (`sequence_length`).
            Position outside of the sequence are not taken into account for computing the loss.

    Outputs: `Tuple` comprising various elements depending on the configuration (config) and inputs:
        **loss**: (`optional`, returned when ``labels`` is provided) ``torch.FloatTensor`` of shape ``(1,)``:
            Total span extraction loss is the sum of a Cross-Entropy for the start and end positions.
        **start_scores**: ``torch.FloatTensor`` of shape ``(batch_size, sequence_length,)``
            Span-start scores (before SoftMax).
        **end_scores**: ``torch.FloatTensor`` of shape ``(batch_size, sequence_length,)``
            Span-end scores (before SoftMax).
        **hidden_states**: (`optional`, returned when ``config.output_hidden_states=True``)
            list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
            of shape ``(batch_size, sequence_length, hidden_size)``:
            Hidden-states of the model at the output of each layer plus the initial embedding outputs.
        **attentions**: (`optional`, returned when ``config.output_attentions=True``)
            list of ``torch.FloatTensor`` (one for each layer) of shape ``(batch_size, num_heads, sequence_length, sequence_length)``:
            Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

    Examples::

        tokenizer = XxxTokenizer.from_pretrained('xxx-base-uncased')
        model = XxxForQuestionAnswering.from_pretrained('xxx-large-uncased-whole-word-masking-finetuned-squad')
        question, text = "Who was Jim Henson?", "Jim Henson was a nice puppet"
        input_text = "[CLS] " + question + " [SEP] " + text + " [SEP]"
        input_ids = tokenizer.encode(input_text)
        token_type_ids = [0 if i <= input_ids.index(102) else 1 for i in range(len(input_ids))]
        start_scores, end_scores = model(torch.tensor([input_ids]), token_type_ids=torch.tensor([token_type_ids]))
        all_tokens = tokenizer.convert_ids_to_tokens(input_ids)
        print(' '.join(all_tokens[torch.argmax(start_scores) : torch.argmax(end_scores)+1]))
        # a nice puppet


    """

    def __init__(self, config):
        super(XxxForQuestionAnswering, self).__init__(config)
        self.num_labels = config.num_labels

        self.transformer = XxxModel(config)
        self.qa_outputs = nn.Linear(config.hidden_size, config.num_labels)

        self.init_weights()

    def forward(
        self,
        input_ids=None,
        attention_mask=None,
        token_type_ids=None,
        position_ids=None,
        head_mask=None,
        inputs_embeds=None,
        start_positions=None,
        end_positions=None,
    ):

        outputs = self.transformer(
            input_ids,
            attention_mask=attention_mask,
            token_type_ids=token_type_ids,
            position_ids=position_ids,
            head_mask=head_mask,
            inputs_embeds=inputs_embeds,
        )

        sequence_output = outputs[0]

        logits = self.qa_outputs(sequence_output)
        start_logits, end_logits = logits.split(1, dim=-1)
        start_logits = start_logits.squeeze(-1)
        end_logits = end_logits.squeeze(-1)

        outputs = (start_logits, end_logits,) + outputs[2:]
        if start_positions is not None and end_positions is not None:
            # If we are on multi-GPU, split add a dimension
            if len(start_positions.size()) > 1:
                start_positions = start_positions.squeeze(-1)
            if len(end_positions.size()) > 1:
                end_positions = end_positions.squeeze(-1)
            # sometimes the start/end positions are outside our model inputs, we ignore these terms
            ignored_index = start_logits.size(1)
            start_positions.clamp_(0, ignored_index)
            end_positions.clamp_(0, ignored_index)

            loss_fct = CrossEntropyLoss(ignore_index=ignored_index)
            start_loss = loss_fct(start_logits, start_positions)
            end_loss = loss_fct(end_logits, end_positions)
            total_loss = (start_loss + end_loss) / 2
            outputs = (total_loss,) + outputs

        return outputs  # (loss), start_logits, end_logits, (hidden_states), (attentions)