modeling_tf_utils.py

# coding=utf-8
# Copyright 2018 The Google AI Language Team Authors and The HuggingFace Inc. team.
# Copyright (c) 2018, NVIDIA CORPORATION.  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.
"""TF general model utils."""

from __future__ import (absolute_import, division, print_function,
                        unicode_literals)

import logging
import os

import tensorflow as tf

from .configuration_utils import PretrainedConfig
from .file_utils import cached_path, WEIGHTS_NAME, TF_WEIGHTS_NAME

logger = logging.getLogger(__name__)


class TFPreTrainedModel(tf.keras.Model):
    r""" Base class for all TF models.

        :class:`~pytorch_transformers.TFPreTrainedModel` takes care of storing the configuration of the models and handles methods for loading/downloading/saving models
        as well as a few methods commons to all models to (i) resize the input embeddings and (ii) prune heads in the self-attention heads.

        Class attributes (overridden by derived classes):
            - ``config_class``: a class derived from :class:`~pytorch_transformers.PretrainedConfig` to use as configuration class for this model architecture.
            - ``pretrained_model_archive_map``: a python ``dict`` of with `short-cut-names` (string) as keys and `url` (string) of associated pretrained weights as values.
            - ``load_tf_weights``: a python ``method`` for loading a TensorFlow checkpoint in a PyTorch model, taking as arguments:

                - ``model``: an instance of the relevant subclass of :class:`~pytorch_transformers.PreTrainedModel`,
                - ``config``: an instance of the relevant subclass of :class:`~pytorch_transformers.PretrainedConfig`,
                - ``path``: a path (string) to the TensorFlow checkpoint.

            - ``base_model_prefix``: a string indicating the attribute associated to the base model in derived classes of the same architecture adding modules on top of the base model.
    """
    config_class = None
    pretrained_model_archive_map = {}
    load_pt_weights = lambda model, config, path: None
    base_model_prefix = ""

    def __init__(self, config, *inputs, **kwargs):
        super(TFPreTrainedModel, self).__init__()
        if not isinstance(config, PretrainedConfig):
            raise ValueError(
                "Parameter config in `{}(config)` should be an instance of class `PretrainedConfig`. "
                "To create a model from a pretrained model use "
                "`model = {}.from_pretrained(PRETRAINED_MODEL_NAME)`".format(
                    self.__class__.__name__, self.__class__.__name__
                ))
        # Save config in model
        self.config = config

    def _get_resized_embeddings(self, old_embeddings, new_num_tokens=None):
        """ Build a resized Embedding Module from a provided token Embedding Module.
            Increasing the size will add newly initialized vectors at the end
            Reducing the size will remove vectors from the end

        Args:
            new_num_tokens: (`optional`) int
                New number of tokens in the embedding matrix.
                Increasing the size will add newly initialized vectors at the end
                Reducing the size will remove vectors from the end
                If not provided or None: return the provided token Embedding Module.
        Return: ``torch.nn.Embeddings``
            Pointer to the resized Embedding Module or the old Embedding Module if new_num_tokens is None
        """
        raise NotImplementedError

    def _tie_or_clone_weights(self, first_module, second_module):
        """ Tie or clone module weights depending of weither we are using TorchScript or not
        """
        raise NotImplementedError

    def resize_token_embeddings(self, new_num_tokens=None):
        """ Resize input token embeddings matrix of the model if new_num_tokens != config.vocab_size.
        Take care of tying weights embeddings afterwards if the model class has a `tie_weights()` method.

        Arguments:

            new_num_tokens: (`optional`) int:
                New number of tokens in the embedding matrix. Increasing the size will add newly initialized vectors at the end. Reducing the size will remove vectors from the end. 
                If not provided or None: does nothing and just returns a pointer to the input tokens ``torch.nn.Embeddings`` Module of the model.

        Return: ``torch.nn.Embeddings``
            Pointer to the input tokens Embeddings Module of the model
        """
        raise NotImplementedError

    def prune_heads(self, heads_to_prune):
        """ Prunes heads of the base model.

            Arguments:

                heads_to_prune: dict with keys being selected layer indices (`int`) and associated values being the list of heads to prune in said layer (list of `int`).
        """
        raise NotImplementedError

    def save_pretrained(self, save_directory):
        """ Save a model and its configuration file to a directory, so that it
            can be re-loaded using the `:func:`~pytorch_transformers.PreTrainedModel.from_pretrained`` class method.
        """
        raise NotImplementedError

    @classmethod
    def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
        r"""Instantiate a pretrained pytorch model from a pre-trained model configuration.

        The model is set in evaluation mode by default using ``model.eval()`` (Dropout modules are deactivated)
        To train the model, you should first set it back in training mode with ``model.train()``

        The warning ``Weights from XXX not initialized from pretrained model`` means that the weights of XXX do not come pre-trained with the rest of the model.
        It is up to you to train those weights with a downstream fine-tuning task.

        The warning ``Weights from XXX not used in YYY`` means that the layer XXX is not used by YYY, therefore those weights are discarded.

        Parameters:
            pretrained_model_name_or_path: either:

                - a string with the `shortcut name` of a pre-trained model to load from cache or download, e.g.: ``bert-base-uncased``.
                - a path to a `directory` containing model weights saved using :func:`~pytorch_transformers.PreTrainedModel.save_pretrained`, e.g.: ``./my_model_directory/``.
                - a path or url to a `PyTorch state_dict save file` (e.g. `./pt_model/pytorch_model.bin`). In this case, ``from_pt`` should be set to True and a configuration object should be provided as ``config`` argument. This loading path is slower than converting the PyTorch checkpoint in a TensorFlow model using the provided conversion scripts and loading the TensorFlow model afterwards.

            model_args: (`optional`) Sequence of positional arguments:
                All remaning positional arguments will be passed to the underlying model's ``__init__`` method

            config: (`optional`) instance of a class derived from :class:`~pytorch_transformers.PretrainedConfig`:
                Configuration for the model to use instead of an automatically loaded configuation. Configuration can be automatically loaded when:

                - the model is a model provided by the library (loaded with the ``shortcut-name`` string of a pretrained model), or
                - the model was saved using :func:`~pytorch_transformers.PreTrainedModel.save_pretrained` and is reloaded by suppling the save directory.
                - the model is loaded by suppling a local directory as ``pretrained_model_name_or_path`` and a configuration JSON file named `config.json` is found in the directory.

            from_pt: (`optional`) boolean, default False:
                Load the model weights from a PyTorch state_dict save file (see docstring of pretrained_model_name_or_path argument).

            cache_dir: (`optional`) string:
                Path to a directory in which a downloaded pre-trained model
                configuration should be cached if the standard cache should not be used.

            force_download: (`optional`) boolean, default False:
                Force to (re-)download the model weights and configuration files and override the cached versions if they exists.

            proxies: (`optional`) dict, default None:
                A dictionary of proxy servers to use by protocol or endpoint, e.g.: {'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}.
                The proxies are used on each request.

            output_loading_info: (`optional`) boolean:
                Set to ``True`` to also return a dictionnary containing missing keys, unexpected keys and error messages.

            kwargs: (`optional`) Remaining dictionary of keyword arguments:
                Can be used to update the configuration object (after it being loaded) and initiate the model. (e.g. ``output_attention=True``). Behave differently depending on whether a `config` is provided or automatically loaded:

                - If a configuration is provided with ``config``, ``**kwargs`` will be directly passed to the underlying model's ``__init__`` method (we assume all relevant updates to the configuration have already been done)
                - If a configuration is not provided, ``kwargs`` will be first passed to the configuration class initialization function (:func:`~pytorch_transformers.PretrainedConfig.from_pretrained`). Each key of ``kwargs`` that corresponds to a configuration attribute will be used to override said attribute with the supplied ``kwargs`` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's ``__init__`` function.

        Examples::

            model = BertModel.from_pretrained('bert-base-uncased')    # Download model and configuration from S3 and cache.
            model = BertModel.from_pretrained('./test/saved_model/')  # E.g. model was saved using `save_pretrained('./test/saved_model/')`
            model = BertModel.from_pretrained('bert-base-uncased', output_attention=True)  # Update configuration during loading
            assert model.config.output_attention == True
            # Loading from a TF checkpoint file instead of a PyTorch model (slower)
            config = BertConfig.from_json_file('./tf_model/my_tf_model_config.json')
            model = BertModel.from_pretrained('./tf_model/my_tf_checkpoint.ckpt.index', from_pt=True, config=config)

        """
        config = kwargs.pop('config', None)
        cache_dir = kwargs.pop('cache_dir', None)
        from_pt = kwargs.pop('from_pt', False)
        force_download = kwargs.pop('force_download', False)
        proxies = kwargs.pop('proxies', None)
        output_loading_info = kwargs.pop('output_loading_info', False)

        # Load config
        if config is None:
            config, model_kwargs = cls.config_class.from_pretrained(
                pretrained_model_name_or_path, *model_args,
                cache_dir=cache_dir, return_unused_kwargs=True,
                force_download=force_download,
                **kwargs
            )
        else:
            model_kwargs = kwargs

        # Load model
        if pretrained_model_name_or_path in cls.pretrained_model_archive_map:
            archive_file = cls.pretrained_model_archive_map[pretrained_model_name_or_path]
        elif os.path.isdir(pretrained_model_name_or_path):
            if from_pt:
                # Load from a PyTorch checkpoint
                archive_file = os.path.join(pretrained_model_name_or_path, WEIGHTS_NAME)
            else:
                archive_file = os.path.join(pretrained_model_name_or_path, TF_WEIGHTS_NAME)
        else:
            archive_file = pretrained_model_name_or_path
        # redirect to the cache, if necessary
        try:
            resolved_archive_file = cached_path(archive_file, cache_dir=cache_dir, force_download=force_download, proxies=proxies)
        except EnvironmentError:
            if pretrained_model_name_or_path in cls.pretrained_model_archive_map:
                logger.error(
                    "Couldn't reach server at '{}' to download pretrained weights.".format(
                        archive_file))
            else:
                logger.error(
                    "Model name '{}' was not found in model name list ({}). "
                    "We assumed '{}' was a path or url but couldn't find any file "
                    "associated to this path or url.".format(
                        pretrained_model_name_or_path,
                        ', '.join(cls.pretrained_model_archive_map.keys()),
                        archive_file))
            return None
        if resolved_archive_file == archive_file:
            logger.info("loading weights file {}".format(archive_file))
        else:
            logger.info("loading weights file {} from cache at {}".format(
                archive_file, resolved_archive_file))

        # Instantiate model.
        model = cls(config, *model_args, **model_kwargs)

        if from_pt:
            # Load from a PyTorch checkpoint
            return cls.load_pt_weights(model, config, resolved_archive_file)

        inputs = tf.constant([[7, 6, 0, 0, 1], [1, 2, 3, 0, 0], [0, 0, 0, 4, 5]])
        ret = model(inputs, training=False)  # build the network with dummy inputs

        # 'by_name' allow us to do transfer learning by skipping/adding layers
        # see https://github.com/tensorflow/tensorflow/blob/00fad90125b18b80fe054de1055770cfb8fe4ba3/tensorflow/python/keras/engine/network.py#L1339-L1357
        model.load_weights(resolved_archive_file, by_name=True)

        ret = model(inputs, training=False)  # Make sure restore ops are run

        # if hasattr(model, 'tie_weights'):
        #     model.tie_weights()  # TODO make sure word embedding weights are still tied

        if output_loading_info:
            loading_info = {"missing_keys": missing_keys, "unexpected_keys": unexpected_keys, "error_msgs": error_msgs}
            return model, loading_info

        return model