# Copyright (c) 2023, Teriks
#
# dgenerate is distributed under the following BSD 3-Clause License
#
# Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in
# the documentation and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote products derived
# from this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
# HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
# ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import os.path
import diffusers
import huggingface_hub
import safetensors
import safetensors.torch
import torch
import dgenerate.messages as _messages
import dgenerate.pipelinewrapper.hfutil as _hfutil
import dgenerate.textprocessing as _textprocessing
import dgenerate.types as _types
from dgenerate.pipelinewrapper.uris import exceptions as _exceptions
_textual_inversion_uri_parser = _textprocessing.ConceptUriParser('Textual Inversion',
['token', 'revision', 'subfolder', 'weight-name'])
def _load_textual_inversion_state_dict(pretrained_model_name_or_path, **kwargs):
from diffusers.utils.hub_utils import _get_model_file
text_inversion_name = "learned_embeds.bin"
text_inversion_name_safe = "learned_embeds.safetensors"
cache_dir = kwargs.pop("cache_dir", None)
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", None)
token = kwargs.pop("token", None)
revision = kwargs.pop("revision", None)
subfolder = kwargs.pop("subfolder", None)
weight_name = kwargs.pop("weight_name", None)
use_safetensors = kwargs.pop("use_safetensors", None)
allow_pickle = False
if use_safetensors is None:
use_safetensors = True
allow_pickle = True
user_agent = {
"file_type": "text_inversion",
"framework": "pytorch",
}
# 3.1. Load textual inversion file
state_dict = None
model_file = None
# Let's first try to load .safetensors weights
if (use_safetensors and weight_name is None) or (
weight_name is not None and weight_name.endswith(".safetensors")
):
try:
model_file = _get_model_file(
pretrained_model_name_or_path,
weights_name=weight_name or text_inversion_name_safe,
cache_dir=cache_dir,
force_download=force_download,
resume_download=resume_download,
proxies=proxies,
local_files_only=local_files_only,
token=token,
revision=revision,
subfolder=subfolder,
user_agent=user_agent,
)
state_dict = safetensors.torch.load_file(model_file, device="cpu")
except Exception as e:
if not allow_pickle:
raise e
model_file = None
if model_file is None:
model_file = _get_model_file(
pretrained_model_name_or_path,
weights_name=weight_name or text_inversion_name,
cache_dir=cache_dir,
force_download=force_download,
resume_download=resume_download,
proxies=proxies,
local_files_only=local_files_only,
token=token,
revision=revision,
subfolder=subfolder,
user_agent=user_agent,
)
state_dict = torch.load(model_file, map_location="cpu")
return model_file, state_dict
[docs]
class TextualInversionUri:
"""
Representation of ``--textual-inversions`` uri
"""
@property
def model(self) -> str:
"""
Model path, huggingface slug, file path
"""
return self._model
@property
def revision(self) -> _types.OptionalString:
"""
Model repo revision
"""
return self._revision
@property
def subfolder(self) -> _types.OptionalPath:
"""
Model repo subfolder
"""
return self._subfolder
@property
def weight_name(self) -> _types.OptionalName:
"""
Model weight-name
"""
return self._weight_name
@property
def token(self) -> _types.OptionalString:
"""
Prompt keyword
"""
return self._token
[docs]
def __init__(self,
model: str,
token: str | None = None,
revision: _types.OptionalString = None,
subfolder: _types.OptionalPath = None,
weight_name: _types.OptionalName = None):
self._token = token
self._model = model
self._revision = revision
self._subfolder = subfolder
self._weight_name = weight_name
def __str__(self):
return f'{self.__class__.__name__}({str(_types.get_public_attributes(self))})'
def __repr__(self):
return str(self)
[docs]
def load_on_pipeline(self,
pipeline: diffusers.DiffusionPipeline,
use_auth_token: _types.OptionalString = None,
local_files_only: bool = False):
"""
Load Textual Inversion weights on to a pipeline using this URI
:param pipeline: :py:class:`diffusers.DiffusionPipeline`
:param use_auth_token: optional huggingface auth token.
:param local_files_only: avoid downloading files and only look for cached files
when the model path is a huggingface slug
:raises ModelNotFoundError: If the model could not be found.
"""
try:
self._load_on_pipeline(pipeline=pipeline,
use_auth_token=use_auth_token,
local_files_only=local_files_only)
except (huggingface_hub.utils.HFValidationError,
huggingface_hub.utils.HfHubHTTPError) as e:
raise _hfutil.ModelNotFoundError(e)
except Exception as e:
raise _exceptions.TextualInversionUriLoadError(
f'error loading textual inversion "{self.model}": {e}')
def _load_on_pipeline(self,
pipeline: diffusers.DiffusionPipeline,
use_auth_token: _types.OptionalString = None,
local_files_only: bool = False):
if hasattr(pipeline, 'load_textual_inversion'):
debug_args = {k: v for k, v in locals().items() if k not in {'self', 'pipeline'}}
_messages.debug_log('pipeline.load_textual_inversion(' +
str(_types.get_public_attributes(self) | debug_args) + ')')
model_path = _hfutil.download_non_hf_model(self.model)
# this is tricky because there is stupidly a positional argument named 'token'
# as well as an accepted kwargs value with the key 'token'
old_token = os.environ.get('HF_TOKEN', None)
if use_auth_token is not None:
os.environ['HF_TOKEN'] = use_auth_token
try:
if pipeline.__class__.__name__.startswith('StableDiffusionXL'):
filename, dicts = _load_textual_inversion_state_dict(
model_path,
revision=self.revision,
subfolder=self.subfolder,
weight_name=self.weight_name,
local_files_only=local_files_only
)
if 'clip_l' not in dicts or 'clip_g' not in dicts:
raise RuntimeError(
'clip_l or clip_g not found in SDXL textual '
f'inversion model "{self.model}" state dict, '
'unsupported model format.')
# token is the file name (no extension) with spaces
# replaced by underscores when the user does not provide
# a prompt token
token = os.path.splitext(
os.path.basename(filename))[0].replace(' ', '_') \
if self.token is None else self.token
pipeline.load_textual_inversion(dicts['clip_l'],
token=token,
text_encoder=pipeline.text_encoder,
tokenizer=pipeline.tokenizer)
pipeline.load_textual_inversion(dicts['clip_g'],
token=token,
text_encoder=pipeline.text_encoder_2,
tokenizer=pipeline.tokenizer_2)
else:
pipeline.load_textual_inversion(model_path,
token=self.token,
revision=self.revision,
subfolder=self.subfolder,
weight_name=self.weight_name,
local_files_only=local_files_only)
finally:
if old_token is not None:
os.environ['HF_TOKEN'] = old_token
_messages.debug_log(f'Added Textual Inversion: "{self}" to pipeline: "{pipeline.__class__.__name__}"')
else:
raise RuntimeError(f'Pipeline: {pipeline.__class__.__name__} '
f'does not support loading textual inversions.')
[docs]
@staticmethod
def parse(uri: _types.Uri) -> 'TextualInversionUri':
"""
Parse a ``--textual-inversions`` uri and return an object representing its constituents
:param uri: string with ``--textual-inversions`` uri syntax
:raise InvalidTextualInversionUriError:
:return: :py:class:`.TextualInversionPath`
"""
try:
r = _textual_inversion_uri_parser.parse(uri)
return TextualInversionUri(model=r.concept,
token=r.args.get('token', None),
weight_name=r.args.get('weight-name', None),
revision=r.args.get('revision', None),
subfolder=r.args.get('subfolder', None))
except _textprocessing.ConceptUriParseError as e:
raise _exceptions.InvalidTextualInversionUriError(e)