FM / transformers

/usr/local/lib/python3.6/site-packages/transformers/

/usr/local/lib/python3.6/site-packages/transformers

Name	Size	Mode	Actions
benchmark/	-	0755	rm
commands/	-	0755	rm
data/	-	0755	rm
models/	-	0755	rm
onnx/	-	0755	rm
pipelines/	-	0755	rm
sagemaker/	-	0755	rm
utils/	-	0755	rm
__pycache__/	-	0755	rm
activations.py	6612	0644	edit dl rm
activations_tf.py	4268	0644	edit dl rm
configuration_utils.py	46765	0644	edit dl rm
convert_graph_to_onnx.py	19575	0644	edit dl rm
convert_pytorch_checkpoint_to_tf2.py	16640	0644	edit dl rm
convert_slow_tokenizer.py	37626	0644	edit dl rm
convert_slow_tokenizers_checkpoints_to_fast.py	4954	0644	edit dl rm
convert_tf_hub_seq_to_seq_bert_to_pytorch.py	2899	0644	edit dl rm
debug_utils.py	12887	0644	edit dl rm
deepspeed.py	18883	0644	edit dl rm
dependency_versions_check.py	1767	0644	edit dl rm
dependency_versions_table.py	2390	0644	edit dl rm
dynamic_module_utils.py	19306	0644	edit dl rm
feature_extraction_sequence_utils.py	18074	0644	edit dl rm
feature_extraction_utils.py	26768	0644	edit dl rm
file_utils.py	3930	0644	edit dl rm
generation_beam_constraints.py	19078	0644	edit dl rm
generation_beam_search.py	40145	0644	edit dl rm
generation_flax_logits_process.py	10879	0644	edit dl rm
generation_flax_utils.py	38708	0644	edit dl rm
generation_logits_process.py	30336	0644	edit dl rm
generation_stopping_criteria.py	5509	0644	edit dl rm
generation_tf_logits_process.py	17503	0644	edit dl rm
generation_tf_utils.py	130671	0644	edit dl rm
generation_utils.py	177118	0644	edit dl rm
hf_argparser.py	11901	0644	edit dl rm
image_utils.py	12860	0644	edit dl rm
integrations.py	40203	0644	edit dl rm
keras_callbacks.py	18902	0644	edit dl rm
modelcard.py	35278	0644	edit dl rm
modeling_flax_outputs.py	35514	0644	edit dl rm
modeling_flax_pytorch_utils.py	12296	0644	edit dl rm
modeling_flax_utils.py	37661	0644	edit dl rm
modeling_outputs.py	60327	0644	edit dl rm
modeling_tf_outputs.py	46107	0644	edit dl rm
modeling_tf_pytorch_utils.py	19916	0644	edit dl rm
modeling_tf_utils.py	100832	0644	edit dl rm
modeling_utils.py	137800	0644	edit dl rm
optimization.py	27756	0644	edit dl rm
optimization_tf.py	15722	0644	edit dl rm
processing_utils.py	10559	0644	edit dl rm
py.typed	1	0644	edit dl rm
pytorch_utils.py	1649	0644	edit dl rm
testing_utils.py	48292	0644	edit dl rm
tf_utils.py	1568	0644	edit dl rm
tokenization_utils.py	39867	0644	edit dl rm
tokenization_utils_base.py	170159	0644	edit dl rm
tokenization_utils_fast.py	32909	0644	edit dl rm
trainer.py	143103	0644	edit dl rm
trainer_callback.py	23199	0644	edit dl rm
trainer_pt_utils.py	44764	0644	edit dl rm
trainer_seq2seq.py	10396	0644	edit dl rm
trainer_tf.py	34564	0644	edit dl rm
trainer_utils.py	18509	0644	edit dl rm
training_args.py	68042	0644	edit dl rm
training_args_seq2seq.py	2829	0644	edit dl rm
training_args_tf.py	14267	0644	edit dl rm
__init__.py	171952	0644	edit dl rm

Edit: /usr/local/lib/python3.6/site-packages/transformers/modeling_flax_utils.py (37661B)

# coding=utf-8
# Copyright 2021 The Google Flax Team Authors and The HuggingFace Inc. team.
#
# 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.

import os
from functools import partial
from pickle import UnpicklingError
from typing import Any, Dict, Set, Tuple, Union

import flax.linen as nn
import jax
import jax.numpy as jnp
import msgpack.exceptions
from flax.core.frozen_dict import FrozenDict, unfreeze
from flax.serialization import from_bytes, to_bytes
from flax.traverse_util import flatten_dict, unflatten_dict
from jax.random import PRNGKey
from requests import HTTPError

from .configuration_utils import PretrainedConfig
from .dynamic_module_utils import custom_object_save
from .generation_flax_utils import FlaxGenerationMixin
from .modeling_flax_pytorch_utils import load_pytorch_checkpoint_in_flax_state_dict
from .utils import (
    FLAX_WEIGHTS_NAME,
    HUGGINGFACE_CO_RESOLVE_ENDPOINT,
    WEIGHTS_NAME,
    EntryNotFoundError,
    PushToHubMixin,
    RepositoryNotFoundError,
    RevisionNotFoundError,
    add_code_sample_docstrings,
    add_start_docstrings_to_model_forward,
    cached_path,
    copy_func,
    has_file,
    hf_bucket_url,
    is_offline_mode,
    is_remote_url,
    logging,
    replace_return_docstrings,
)

logger = logging.get_logger(__name__)

def quick_gelu(x):
    return x * jax.nn.sigmoid(1.702 * x)

ACT2FN = {
    "gelu": partial(nn.gelu, approximate=False),
    "relu": nn.relu,
    "silu": nn.swish,
    "swish": nn.swish,
    "gelu_new": partial(nn.gelu, approximate=True),
    "quick_gelu": quick_gelu,
}

class FlaxPreTrainedModel(PushToHubMixin, FlaxGenerationMixin):
    r"""
    Base class for all models.

[`FlaxPreTrainedModel`] takes care of storing the configuration of the models and handles methods for loading,
    downloading and saving models.

Class attributes (overridden by derived classes):

- **config_class** ([`PretrainedConfig`]) -- A subclass of [`PretrainedConfig`] to use as configuration class
          for this model architecture.
        - **base_model_prefix** (`str`) -- 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.
        - **main_input_name** (`str`) -- The name of the principal input to the model (often `input_ids` for NLP
          models, `pixel_values` for vision models and `input_values` for speech models).
    """
    config_class = None
    base_model_prefix = ""
    main_input_name = "input_ids"
    _auto_class = None

def __init__(
        self,
        config: PretrainedConfig,
        module: nn.Module,
        input_shape: Tuple = (1, 1),
        seed: int = 0,
        dtype: jnp.dtype = jnp.float32,
    ):
        if config is None:
            raise ValueError("config cannot be None")

if module is None:
            raise ValueError("module cannot be None")

# Those are private to be exposed as typed property on derived classes.
        self._config = config
        self._module = module

# Those are public as their type is generic to every derived classes.
        self.key = PRNGKey(seed)
        self.dtype = dtype

# randomly initialized parameters
        random_params = self.init_weights(self.key, input_shape)

# save required_params as set
        self._required_params = set(flatten_dict(unfreeze(random_params)).keys())
        self.params = random_params

def init_weights(self, rng: jax.random.PRNGKey, input_shape: Tuple) -> Dict:
        raise NotImplementedError(f"init method has to be implemented for {self}")

@classmethod
    def _from_config(cls, config, **kwargs):
        """
        All context managers that the model should be initialized under go here.
        """
        return cls(config, **kwargs)

@property
    def framework(self) -> str:
        """
        :str: Identifies that this is a Flax model.
        """
        return "flax"

@property
    def config(self) -> PretrainedConfig:
        return self._config

@property
    def module(self) -> nn.Module:
        return self._module

@property
    def params(self) -> Union[Dict, FrozenDict]:
        return self._params

@property
    def required_params(self) -> Set:
        return self._required_params

@params.setter
    def params(self, params: Union[Dict, FrozenDict]):
        if isinstance(params, FrozenDict):
            params = unfreeze(params)
        param_keys = set(flatten_dict(params).keys())
        if len(self.required_params - param_keys) > 0:
            raise ValueError(
                "Some parameters are missing. Make sure that `params` include the following "
                f"parameters {self.required_params - param_keys}"
            )
        self._params = params

def _cast_floating_to(self, params: Union[Dict, FrozenDict], dtype: jnp.dtype, mask: Any = None) -> Any:
        """
        Helper method to cast floating-point values of given parameter `PyTree` to given `dtype`.
        """

# taken from https://github.com/deepmind/jmp/blob/3a8318abc3292be38582794dbf7b094e6583b192/jmp/_src/policy.py#L27
        def conditional_cast(param):
            if isinstance(param, jnp.ndarray) and jnp.issubdtype(param.dtype, jnp.floating):
                param = param.astype(dtype)
            return param

if mask is None:
            return jax.tree_map(conditional_cast, params)

flat_params = flatten_dict(params)
        flat_mask, _ = jax.tree_flatten(mask)

for masked, key in zip(flat_mask, flat_params.keys()):
            if masked:
                param = flat_params[key]
                flat_params[key] = conditional_cast(param)

return unflatten_dict(flat_params)

def to_bf16(self, params: Union[Dict, FrozenDict], mask: Any = None):
        r"""
        Cast the floating-point `params` to `jax.numpy.bfloat16`. This returns a new `params` tree and does not cast
        the `params` in place.

This method can be used on TPU to explicitly convert the model parameters to bfloat16 precision to do full
        half-precision training or to save weights in bfloat16 for inference in order to save memory and improve speed.

Arguments:
            params (`Union[Dict, FrozenDict]`):
                A `PyTree` of model parameters.
            mask (`Union[Dict, FrozenDict]`):
                A `PyTree` with same structure as the `params` tree. The leaves should be booleans, `True` for params
                you want to cast, and should be `False` for those you want to skip.

Examples:

```python
        >>> from transformers import FlaxBertModel

>>> # load model
        >>> model = FlaxBertModel.from_pretrained("bert-base-cased")
        >>> # By default, the model parameters will be in fp32 precision, to cast these to bfloat16 precision
        >>> model.params = model.to_bf16(model.params)
        >>> # If you want don't want to cast certain parameters (for example layer norm bias and scale)
        >>> # then pass the mask as follows
        >>> from flax import traverse_util

>>> model = FlaxBertModel.from_pretrained("bert-base-cased")
        >>> flat_params = traverse_util.flatten_dict(model.params)
        >>> mask = {
        ...     path: (path[-2] != ("LayerNorm", "bias") and path[-2:] != ("LayerNorm", "scale"))
        ...     for path in flat_params
        ... }
        >>> mask = traverse_util.unflatten_dict(mask)
        >>> model.params = model.to_bf16(model.params, mask)
        ```"""
        return self._cast_floating_to(params, jnp.bfloat16, mask)

def to_fp32(self, params: Union[Dict, FrozenDict], mask: Any = None):
        r"""
        Cast the floating-point `parmas` to `jax.numpy.float32`. This method can be used to explicitly convert the
        model parameters to fp32 precision. This returns a new `params` tree and does not cast the `params` in place.

Examples:

```python
        >>> from transformers import FlaxBertModel

>>> # Download model and configuration from huggingface.co
        >>> model = FlaxBertModel.from_pretrained("bert-base-cased")
        >>> # By default, the model params will be in fp32, to illustrate the use of this method,
        >>> # we'll first cast to fp16 and back to fp32
        >>> model.params = model.to_f16(model.params)
        >>> # now cast back to fp32
        >>> model.params = model.to_fp32(model.params)
        ```"""
        return self._cast_floating_to(params, jnp.float32, mask)

def to_fp16(self, params: Union[Dict, FrozenDict], mask: Any = None):
        r"""
        Cast the floating-point `parmas` to `jax.numpy.float16`. This returns a new `params` tree and does not cast the
        `params` in place.

This method can be used on GPU to explicitly convert the model parameters to float16 precision to do full
        half-precision training or to save weights in float16 for inference in order to save memory and improve speed.

Examples:

```python
        >>> from transformers import FlaxBertModel

>>> # load model
        >>> model = FlaxBertModel.from_pretrained("bert-base-cased")
        >>> # By default, the model params will be in fp32, to cast these to float16
        >>> model.params = model.to_fp16(model.params)
        >>> # If you want don't want to cast certain parameters (for example layer norm bias and scale)
        >>> # then pass the mask as follows
        >>> from flax import traverse_util

>>> model = FlaxBertModel.from_pretrained("bert-base-cased")
        >>> flat_params = traverse_util.flatten_dict(model.params)
        >>> mask = {
        ...     path: (path[-2] != ("LayerNorm", "bias") and path[-2:] != ("LayerNorm", "scale"))
        ...     for path in flat_params
        ... }
        >>> mask = traverse_util.unflatten_dict(mask)
        >>> model.params = model.to_fp16(model.params, mask)
        ```"""
        return self._cast_floating_to(params, jnp.float16, mask)

@classmethod
    def from_pretrained(
        cls,
        pretrained_model_name_or_path: Union[str, os.PathLike],
        dtype: jnp.dtype = jnp.float32,
        *model_args,
        **kwargs
    ):

r"""
        Instantiate a pretrained flax model from a pre-trained model configuration.

The warning *Weights from XXX not initialized from pretrained model* means that the weights of XXX do not come
        pretrained 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 (`str` or `os.PathLike`):
                Can be either:

- A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co.
                      Valid model ids can be located at the root-level, like `bert-base-uncased`, or namespaced under a
                      user or organization name, like `dbmdz/bert-base-german-cased`.
                    - A path to a *directory* containing model weights saved using
                      [`~FlaxPreTrainedModel.save_pretrained`], e.g., `./my_model_directory/`.
                    - A path or url to a *pt index checkpoint file* (e.g, `./tf_model/model.ckpt.index`). In this case,
                      `from_pt` should be set to `True`.
            dtype (`jax.numpy.dtype`, *optional*, defaults to `jax.numpy.float32`):
                The data type of the computation. Can be one of `jax.numpy.float32`, `jax.numpy.float16` (on GPUs) and
                `jax.numpy.bfloat16` (on TPUs).

This can be used to enable mixed-precision training or half-precision inference on GPUs or TPUs. If
                specified all the computation will be performed with the given `dtype`.

**Note that this only specifies the dtype of the computation and does not influence the dtype of model
                parameters.**

If you wish to change the dtype of the model parameters, see [`~FlaxPreTrainedModel.to_fp16`] and
                [`~FlaxPreTrainedModel.to_bf16`].
            model_args (sequence of positional arguments, *optional*):
                All remaining positional arguments will be passed to the underlying model's `__init__` method.
            config (`Union[PretrainedConfig, str, os.PathLike]`, *optional*):
                Can be either:

- an instance of a class derived from [`PretrainedConfig`],
                    - a string or path valid as input to [`~PretrainedConfig.from_pretrained`].

Configuration for the model to use instead of an automatically loaded configuration. Configuration can
                be automatically loaded when:

- The model is a model provided by the library (loaded with the *model id* string of a pretrained
                      model).
                    - The model was saved using [`~PreTrainedModel.save_pretrained`] and is reloaded by supplying the
                      save directory.
                    - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a
                      configuration JSON file named *config.json* is found in the directory.
            cache_dir (`Union[str, os.PathLike]`, *optional*):
                Path to a directory in which a downloaded pretrained model configuration should be cached if the
                standard cache should not be used.
            from_pt (`bool`, *optional*, defaults to `False`):
                Load the model weights from a PyTorch checkpoint save file (see docstring of
                `pretrained_model_name_or_path` argument).
            ignore_mismatched_sizes (`bool`, *optional*, defaults to `False`):
                Whether or not to raise an error if some of the weights from the checkpoint do not have the same size
                as the weights of the model (if for instance, you are instantiating a model with 10 labels from a
                checkpoint with 3 labels).
            force_download (`bool`, *optional*, defaults to `False`):
                Whether or not to force the (re-)download of the model weights and configuration files, overriding the
                cached versions if they exist.
            resume_download (`bool`, *optional*, defaults to `False`):
                Whether or not to delete incompletely received files. Will attempt to resume the download if such a
                file exists.
            proxies (`Dict[str, str]`, *optional*):
                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.
            local_files_only(`bool`, *optional*, defaults to `False`):
                Whether or not to only look at local files (i.e., do not try to download the model).
            revision (`str`, *optional*, defaults to `"main"`):
                The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a
                git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any
                identifier allowed by git.
            kwargs (remaining dictionary of keyword arguments, *optional*):
                Can be used to update the configuration object (after it being loaded) and initiate the model (e.g.,
                `output_attentions=True`). Behaves 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 ([`~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:

```python
        >>> from transformers import BertConfig, FlaxBertModel

>>> # Download model and configuration from huggingface.co and cache.
        >>> model = FlaxBertModel.from_pretrained("bert-base-cased")
        >>> # Model was saved using *save_pretrained('./test/saved_model/')* (for example purposes, not runnable).
        >>> model = FlaxBertModel.from_pretrained("./test/saved_model/")
        >>> # Loading from a PyTorch checkpoint file instead of a PyTorch model (slower, for example purposes, not runnable).
        >>> config = BertConfig.from_json_file("./pt_model/config.json")
        >>> model = FlaxBertModel.from_pretrained("./pt_model/pytorch_model.bin", from_pt=True, config=config)
        ```"""
        config = kwargs.pop("config", None)
        cache_dir = kwargs.pop("cache_dir", None)
        from_pt = kwargs.pop("from_pt", False)
        ignore_mismatched_sizes = kwargs.pop("ignore_mismatched_sizes", False)
        force_download = kwargs.pop("force_download", False)
        resume_download = kwargs.pop("resume_download", False)
        proxies = kwargs.pop("proxies", None)
        local_files_only = kwargs.pop("local_files_only", False)
        use_auth_token = kwargs.pop("use_auth_token", None)
        revision = kwargs.pop("revision", None)
        from_pipeline = kwargs.pop("_from_pipeline", None)
        from_auto_class = kwargs.pop("_from_auto", False)

user_agent = {"file_type": "model", "framework": "flax", "from_auto_class": from_auto_class}
        if from_pipeline is not None:
            user_agent["using_pipeline"] = from_pipeline

if is_offline_mode() and not local_files_only:
            logger.info("Offline mode: forcing local_files_only=True")
            local_files_only = True

# Load config if we don't provide a configuration
        if not isinstance(config, PretrainedConfig):
            config_path = config if config is not None else pretrained_model_name_or_path
            config, model_kwargs = cls.config_class.from_pretrained(
                config_path,
                cache_dir=cache_dir,
                return_unused_kwargs=True,
                force_download=force_download,
                resume_download=resume_download,
                proxies=proxies,
                local_files_only=local_files_only,
                use_auth_token=use_auth_token,
                revision=revision,
                _from_auto=from_auto_class,
                _from_pipeline=from_pipeline,
                **kwargs,
            )
        else:
            model_kwargs = kwargs

# Add the dtype to model_kwargs
        model_kwargs["dtype"] = dtype

# Load model
        if pretrained_model_name_or_path is not None:
            if os.path.isdir(pretrained_model_name_or_path):
                if from_pt and os.path.isfile(os.path.join(pretrained_model_name_or_path, WEIGHTS_NAME)):
                    # Load from a PyTorch checkpoint
                    archive_file = os.path.join(pretrained_model_name_or_path, WEIGHTS_NAME)
                elif os.path.isfile(os.path.join(pretrained_model_name_or_path, FLAX_WEIGHTS_NAME)):
                    # Load from a Flax checkpoint
                    archive_file = os.path.join(pretrained_model_name_or_path, FLAX_WEIGHTS_NAME)
                # At this stage we don't have a weight file so we will raise an error.
                elif os.path.join(pretrained_model_name_or_path, WEIGHTS_NAME):
                    raise EnvironmentError(
                        f"Error no file named {FLAX_WEIGHTS_NAME} found in directory {pretrained_model_name_or_path} "
                        "but there is a file for PyTorch weights. Use `from_pt=True` to load this model from those "
                        "weights."
                    )
                else:
                    raise EnvironmentError(
                        f"Error no file named {FLAX_WEIGHTS_NAME} or {WEIGHTS_NAME} found in directory "
                        f"{pretrained_model_name_or_path}."
                    )
            elif os.path.isfile(pretrained_model_name_or_path) or is_remote_url(pretrained_model_name_or_path):
                archive_file = pretrained_model_name_or_path
            else:
                filename = WEIGHTS_NAME if from_pt else FLAX_WEIGHTS_NAME
                archive_file = hf_bucket_url(
                    pretrained_model_name_or_path,
                    filename=filename,
                    revision=revision,
                )

# redirect to the cache, if necessary
            try:
                resolved_archive_file = cached_path(
                    archive_file,
                    cache_dir=cache_dir,
                    force_download=force_download,
                    proxies=proxies,
                    resume_download=resume_download,
                    local_files_only=local_files_only,
                    use_auth_token=use_auth_token,
                    user_agent=user_agent,
                )

except RepositoryNotFoundError:
                raise EnvironmentError(
                    f"{pretrained_model_name_or_path} is not a local folder and is not a valid model identifier "
                    "listed on 'https://huggingface.co/models'\nIf this is a private repository, make sure to pass a "
                    "token having permission to this repo with `use_auth_token` or log in with `huggingface-cli "
                    "login` and pass `use_auth_token=True`."
                )
            except RevisionNotFoundError:
                raise EnvironmentError(
                    f"{revision} is not a valid git identifier (branch name, tag name or commit id) that exists for "
                    "this model name. Check the model page at "
                    f"'https://huggingface.co/{pretrained_model_name_or_path}' for available revisions."
                )
            except EntryNotFoundError:
                if filename == FLAX_WEIGHTS_NAME:
                    has_file_kwargs = {"revision": revision, "proxies": proxies, "use_auth_token": use_auth_token}
                    if has_file(pretrained_model_name_or_path, WEIGHTS_NAME, **has_file_kwargs):
                        raise EnvironmentError(
                            f"{pretrained_model_name_or_path} does not appear to have a file named {FLAX_WEIGHTS_NAME} "
                            "but there is a file for PyTorch weights. Use `from_pt=True` to load this model from "
                            "those weights."
                        )
                    else:
                        raise EnvironmentError(
                            f"{pretrained_model_name_or_path} does not appear to have a file named {FLAX_WEIGHTS_NAME} "
                            f"or {WEIGHTS_NAME}."
                        )
                else:
                    raise EnvironmentError(
                        f"{pretrained_model_name_or_path} does not appear to have a file named {filename}."
                    )
            except HTTPError as err:
                raise EnvironmentError(
                    f"There was a specific connection error when trying to load {pretrained_model_name_or_path}:\n"
                    f"{err}"
                )
            except ValueError:
                raise EnvironmentError(
                    f"We couldn't connect to '{HUGGINGFACE_CO_RESOLVE_ENDPOINT}' to load this model, couldn't find it in the cached "
                    f"files and it looks like {pretrained_model_name_or_path} is not the path to a directory "
                    f"containing a file named {FLAX_WEIGHTS_NAME} or {WEIGHTS_NAME}.\n"
                    "Checkout your internet connection or see how to run the library in offline mode at "
                    "'https://huggingface.co/docs/transformers/installation#offline-mode'."
                )
            except EnvironmentError:
                raise EnvironmentError(
                    f"Can't load the model for '{pretrained_model_name_or_path}'. If you were trying to load it from "
                    "'https://huggingface.co/models', make sure you don't have a local directory with the same name. "
                    f"Otherwise, make sure '{pretrained_model_name_or_path}' is the correct path to a directory "
                    f"containing a file named {FLAX_WEIGHTS_NAME} or {WEIGHTS_NAME}."
                )

if resolved_archive_file == archive_file:
                logger.info(f"loading weights file {archive_file}")
            else:
                logger.info(f"loading weights file {archive_file} from cache at {resolved_archive_file}")
        else:
            resolved_archive_file = None

# init random models
        model = cls(config, *model_args, **model_kwargs)

if from_pt:
            state = load_pytorch_checkpoint_in_flax_state_dict(model, resolved_archive_file)
        else:
            with open(resolved_archive_file, "rb") as state_f:
                try:
                    state = from_bytes(cls, state_f.read())
                except (UnpicklingError, msgpack.exceptions.ExtraData) as e:
                    try:
                        with open(resolved_archive_file) as f:
                            if f.read().startswith("version"):
                                raise OSError(
                                    "You seem to have cloned a repository without having git-lfs installed. Please install "
                                    "git-lfs and run `git lfs install` followed by `git lfs pull` in the folder "
                                    "you cloned."
                                )
                            else:
                                raise ValueError from e
                    except (UnicodeDecodeError, ValueError):
                        raise EnvironmentError(f"Unable to convert {archive_file} to Flax deserializable object. ")
            # make sure all arrays are stored as jnp.arrays
            # NOTE: This is to prevent a bug this will be fixed in Flax >= v0.3.4:
            # https://github.com/google/flax/issues/1261
            state = jax.tree_util.tree_map(jnp.array, state)

# if model is base model only use model_prefix key
        if cls.base_model_prefix not in dict(model.params) and cls.base_model_prefix in state:
            state = state[cls.base_model_prefix]

# if model is head model and we are loading weights from base model
        # we initialize new params dict with base_model_prefix
        if cls.base_model_prefix in dict(model.params) and cls.base_model_prefix not in state:
            state = {cls.base_model_prefix: state}

# flatten dicts
        state = flatten_dict(state)

random_state = flatten_dict(unfreeze(model.params))

missing_keys = model.required_params - set(state.keys())
        unexpected_keys = set(state.keys()) - model.required_params

# Mistmatched keys contains tuples key/shape1/shape2 of weights in the checkpoint that have a shape not
        # matching the weights in the model.
        mismatched_keys = []
        for key in state.keys():
            if key in random_state and state[key].shape != random_state[key].shape:
                if ignore_mismatched_sizes:
                    mismatched_keys.append((key, state[key].shape, random_state[key].shape))
                    state[key] = random_state[key]
                else:
                    raise ValueError(
                        f"Trying to load the pretrained weight for {key} failed: checkpoint has shape "
                        f"{state[key].shape} which is incompatible with the model shape {random_state[key].shape}. "
                        "Using `ignore_mismatched_sizes=True` if you really want to load this checkpoint inside this "
                        "model."
                    )

# add missing keys as random parameters
        for missing_key in missing_keys:
            state[missing_key] = random_state[missing_key]

# remove unexpected keys to not be saved again
        for unexpected_key in unexpected_keys:
            del state[unexpected_key]

if len(unexpected_keys) > 0:
            logger.warning(
                f"Some weights of the model checkpoint at {pretrained_model_name_or_path} were not used when "
                f"initializing {model.__class__.__name__}: {unexpected_keys}\n"
                f"- This IS expected if you are initializing {model.__class__.__name__} from the checkpoint of a model trained on another task "
                f"or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).\n"
                f"- This IS NOT expected if you are initializing {model.__class__.__name__} from the checkpoint of a model that you expect "
                f"to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model)."
            )
        else:
            logger.info(f"All model checkpoint weights were used when initializing {model.__class__.__name__}.\n")

if len(missing_keys) > 0:
            logger.warning(
                f"Some weights of {model.__class__.__name__} were not initialized from the model checkpoint at {pretrained_model_name_or_path} "
                f"and are newly initialized: {missing_keys}\n"
                f"You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference."
            )
        elif len(mismatched_keys) == 0:
            logger.info(
                f"All the weights of {model.__class__.__name__} were initialized from the model checkpoint at {pretrained_model_name_or_path}.\n"
                f"If your task is similar to the task the model of the checkpoint was trained on, "
                f"you can already use {model.__class__.__name__} for predictions without further training."
            )
        if len(mismatched_keys) > 0:
            mismatched_warning = "\n".join(
                [
                    f"- {key}: found shape {shape1} in the checkpoint and {shape2} in the model instantiated"
                    for key, shape1, shape2 in mismatched_keys
                ]
            )
            logger.warning(
                f"Some weights of {model.__class__.__name__} were not initialized from the model checkpoint at {pretrained_model_name_or_path} "
                f"and are newly initialized because the shapes did not match:\n{mismatched_warning}\n"
                f"You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference."
            )

# set correct parameters
        model.params = unflatten_dict(state)

return model

def save_pretrained(self, save_directory: Union[str, os.PathLike], params=None, push_to_hub=False, **kwargs):
        """
        Save a model and its configuration file to a directory, so that it can be re-loaded using the
        `[`~FlaxPreTrainedModel.from_pretrained`]` class method

Arguments:
            save_directory (`str` or `os.PathLike`):
                Directory to which to save. Will be created if it doesn't exist.
            push_to_hub (`bool`, *optional*, defaults to `False`):
                Whether or not to push your model to the Hugging Face model hub after saving it.

Using `push_to_hub=True` will synchronize the repository you are pushing to with `save_directory`,
                which requires `save_directory` to be a local clone of the repo you are pushing to if it's an existing
                folder. Pass along `temp_dir=True` to use a temporary directory instead.

</Tip>

kwargs:
                Additional key word arguments passed along to the [`~utils.PushToHubMixin.push_to_hub`] method.
        """
        if os.path.isfile(save_directory):
            logger.error(f"Provided path ({save_directory}) should be a directory, not a file")
            return

if push_to_hub:
            commit_message = kwargs.pop("commit_message", None)
            repo = self._create_or_get_repo(save_directory, **kwargs)

os.makedirs(save_directory, exist_ok=True)

# get abs dir
        save_directory = os.path.abspath(save_directory)
        # save config as well
        self.config.architectures = [self.__class__.__name__[4:]]

# If we have a custom model, we copy the file defining it in the folder and set the attributes so it can be
        # loaded from the Hub.
        if self._auto_class is not None:
            custom_object_save(self, save_directory, config=self.config)

self.config.save_pretrained(save_directory)

# save model
        output_model_file = os.path.join(save_directory, FLAX_WEIGHTS_NAME)
        with open(output_model_file, "wb") as f:
            params = params if params is not None else self.params
            model_bytes = to_bytes(params)
            f.write(model_bytes)

logger.info(f"Model weights saved in {output_model_file}")

if push_to_hub:
            url = self._push_to_hub(repo, commit_message=commit_message)
            logger.info(f"Model pushed to the hub in this commit: {url}")

@classmethod
    def register_for_auto_class(cls, auto_class="FlaxAutoModel"):
        """
        Register this class with a given auto class. This should only be used for custom models as the ones in the
        library are already mapped with an auto class.

This API is experimental and may have some slight breaking changes in the next releases.

</Tip>

Args:
            auto_class (`str` or `type`, *optional*, defaults to `"FlaxAutoModel"`):
                The auto class to register this new model with.
        """
        if not isinstance(auto_class, str):
            auto_class = auto_class.__name__

import transformers.models.auto as auto_module

if not hasattr(auto_module, auto_class):
            raise ValueError(f"{auto_class} is not a valid auto class.")

cls._auto_class = auto_class

# To update the docstring, we need to copy the method, otherwise we change the original docstring.
FlaxPreTrainedModel.push_to_hub = copy_func(FlaxPreTrainedModel.push_to_hub)
FlaxPreTrainedModel.push_to_hub.__doc__ = FlaxPreTrainedModel.push_to_hub.__doc__.format(
    object="model", object_class="FlaxAutoModel", object_files="model checkpoint"
)

def overwrite_call_docstring(model_class, docstring):
    # copy __call__ function to be sure docstring is changed only for this function
    model_class.__call__ = copy_func(model_class.__call__)
    # delete existing docstring
    model_class.__call__.__doc__ = None
    # set correct docstring
    model_class.__call__ = add_start_docstrings_to_model_forward(docstring)(model_class.__call__)

def append_call_sample_docstring(model_class, tokenizer_class, checkpoint, output_type, config_class, mask=None):
    model_class.__call__ = copy_func(model_class.__call__)
    model_class.__call__ = add_code_sample_docstrings(
        processor_class=tokenizer_class,
        checkpoint=checkpoint,
        output_type=output_type,
        config_class=config_class,
        model_cls=model_class.__name__,
    )(model_class.__call__)

def append_replace_return_docstrings(model_class, output_type, config_class):
    model_class.__call__ = copy_func(model_class.__call__)
    model_class.__call__ = replace_return_docstrings(
        output_type=output_type,
        config_class=config_class,
    )(model_class.__call__)

cmd: