Models

Generic model classes

The following ORT classes are available for instantiating a base model class without a specific head.

ORTModel

class optimum.onnxruntime.ORTModel

< >

( model: InferenceSession config: PretrainedConfig use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None **kwargs )

Base class for implementing models using ONNX Runtime.

The ORTModel implements generic methods for interacting with the Hugging Face Hub as well as exporting vanilla transformers models to ONNX using optimum.exporters.onnx toolchain.

Class attributes:

Common attributes:

can_generate

< >

( )

Returns whether this model can generate sequences with .generate().

from_pretrained

< >

( model_id: typing.Union[str, pathlib.Path] export: bool = False force_download: bool = False use_auth_token: typing.Union[str, bool, NoneType] = None token: typing.Union[str, bool, NoneType] = None cache_dir: str = '/root/.cache/huggingface/hub' subfolder: str = '' config: typing.Optional[ForwardRef('PretrainedConfig')] = None local_files_only: bool = False provider: str = 'CPUExecutionProvider' session_options: typing.Optional[onnxruntime.capi.onnxruntime_pybind11_state.SessionOptions] = None provider_options: typing.Union[typing.Dict[str, typing.Any], NoneType] = None use_io_binding: typing.Optional[bool] = None **kwargs ) ORTModel

Parameters

  • model_id (Union[str, Path]) — 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 a model saved using ~OptimizedModel.save_pretrained, e.g., ./my_model_directory/.
  • export (bool, defaults to False) — Defines whether the provided model_id needs to be exported to the targeted format.
  • force_download (bool, defaults to True) — Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist.
  • use_auth_token (Optional[Union[bool,str]], defaults to None) — Deprecated. Please use the token argument instead.
  • token (Optional[Union[bool,str]], defaults to None) — The token to use as HTTP bearer authorization for remote files. If True, will use the token generated when running huggingface-cli login (stored in huggingface_hub.constants.HF_TOKEN_PATH).
  • cache_dir (Optional[str], defaults to None) — Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used.
  • subfolder (str, defaults to "") — In case the relevant files are located inside a subfolder of the model repo either locally or on huggingface.co, you can specify the folder name here.
  • config (Optional[transformers.PretrainedConfig], defaults to None) — The model configuration.
  • local_files_only (Optional[bool], defaults to False) — Whether or not to only look at local files (i.e., do not try to download the model).
  • trust_remote_code (bool, defaults to False) — Whether or not to allow for custom code defined on the Hub in their own modeling. This option should only be set to True for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine.
  • revision (Optional[str], defaults to None) — 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.
  • provider (str, defaults to "CPUExecutionProvider") — ONNX Runtime provider to use for loading the model. See https://onnxruntime.ai/docs/execution-providers/ for possible providers.
  • session_options (Optional[onnxruntime.SessionOptions], defaults to None), — ONNX Runtime session options to use for loading the model.
  • provider_options (Optional[Dict[str, Any]], defaults to None) — Provider option dictionaries corresponding to the provider used. See available options for each provider: https://onnxruntime.ai/docs/api/c/group___global.html .
  • use_io_binding (Optional[bool], defaults to None) — Whether to use IOBinding during inference to avoid memory copy between the host and device, or between numpy/torch tensors and ONNX Runtime ORTValue. Defaults to True if the execution provider is CUDAExecutionProvider. For [~onnxruntime.ORTModelForCausalLM], defaults to True on CPUExecutionProvider, in all other cases defaults to False.
  • kwargs (Dict[str, Any]) — Will be passed to the underlying model loading methods.

Parameters for decoder models (ORTModelForCausalLM, ORTModelForSeq2SeqLM, ORTModelForSeq2SeqLM, ORTModelForSpeechSeq2Seq, ORTModelForVision2Seq)

  • use_cache (Optional[bool], defaults to True) — Whether or not past key/values cache should be used. Defaults to True.

Parameters for ORTModelForCausalLM

  • use_merged (Optional[bool], defaults to None) — whether or not to use a single ONNX that handles both the decoding without and with past key values reuse. This option defaults to True if loading from a local repository and a merged decoder is found. When exporting with export=True, defaults to False. This option should be set to True to minimize memory usage.

Returns

ORTModel

The loaded ORTModel model.

Instantiate a pretrained model from a pre-trained model configuration.

load_model

< >

( path: typing.Union[str, pathlib.Path] provider: str = 'CPUExecutionProvider' session_options: typing.Optional[onnxruntime.capi.onnxruntime_pybind11_state.SessionOptions] = None provider_options: typing.Union[typing.Dict[str, typing.Any], NoneType] = None )

Parameters

  • path (Union[str, Path]) — Path of the ONNX model.
  • provider (str, defaults to "CPUExecutionProvider") — ONNX Runtime provider to use for loading the model. See https://onnxruntime.ai/docs/execution-providers/ for possible providers.
  • session_options (Optional[onnxruntime.SessionOptions], defaults to None) — ONNX Runtime session options to use for loading the model.
  • provider_options (Optional[Dict[str, Any]], defaults to None) — Provider option dictionary corresponding to the provider used. See available options for each provider: https://onnxruntime.ai/docs/api/c/group___global.html .

Loads an ONNX Inference session with a given provider. Default provider is CPUExecutionProvider to match the default behaviour in PyTorch/TensorFlow/JAX.

raise_on_numpy_input_io_binding

< >

( use_torch: bool )

Parameters

  • use_torch (bool) — Whether the tensor used during inference are of type torch.Tensor or not.

Raises an error if IO Binding is requested although the tensor used are numpy arrays.

shared_attributes_init

< >

( model: InferenceSession use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None **kwargs )

Initializes attributes that may be shared among several ONNX Runtime inference sesssions.

to

< >

( device: typing.Union[torch.device, str, int] ) ORTModel

Parameters

  • device (torch.device or str or int) — Device ordinal for CPU/GPU supports. Setting this to -1 will leverage CPU, a positive will run the model on the associated CUDA device id. You can pass native torch.device or a str too.

Returns

ORTModel

the model placed on the requested device.

Changes the ONNX Runtime provider according to the device.

Natural Language Processing

The following ORT classes are available for the following natural language processing tasks.

ORTModelForCausalLM

class optimum.onnxruntime.ORTModelForCausalLM

< >

( model: InferenceSession config: PretrainedConfig use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None generation_config: typing.Optional[transformers.generation.configuration_utils.GenerationConfig] = None use_cache: typing.Optional[bool] = None **kwargs )

ONNX model with a causal language modeling head for ONNX Runtime inference. This class officially supports bloom, codegen, falcon, gpt2, gpt_bigcode, gpt_neo, gpt_neox, gptj, llama.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( input_ids: LongTensor attention_mask: typing.Optional[torch.FloatTensor] = None position_ids: typing.Optional[torch.LongTensor] = None past_key_values: typing.Optional[typing.Tuple[typing.Tuple[torch.Tensor]]] = None labels: typing.Optional[torch.LongTensor] = None use_cache_branch: bool = None **kwargs )

Parameters

  • input_ids (torch.LongTensor) — Indices of decoder input sequence tokens in the vocabulary of shape (batch_size, sequence_length).
  • attention_mask (torch.LongTensor) — Mask to avoid performing attention on padding token indices, of shape (batch_size, sequence_length). Mask values selected in [0, 1].
  • past_key_values (tuple(tuple(torch.FloatTensor), *optional*, defaults to None) — Contains the precomputed key and value hidden states of the attention blocks used to speed up decoding. The tuple is of length config.n_layers with each tuple having 2 tensors of shape (batch_size, num_heads, sequence_length, embed_size_per_head).

The ORTModelForCausalLM forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of text generation:

>>> from transformers import AutoTokenizer
>>> from optimum.onnxruntime import ORTModelForCausalLM
>>> import torch

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/gpt2")
>>> model = ORTModelForCausalLM.from_pretrained("optimum/gpt2")

>>> inputs = tokenizer("My name is Arthur and I live in", return_tensors="pt")

>>> gen_tokens = model.generate(**inputs,do_sample=True,temperature=0.9, min_length=20,max_length=20)
>>> tokenizer.batch_decode(gen_tokens)

Example using transformers.pipelines:

>>> from transformers import AutoTokenizer, pipeline
>>> from optimum.onnxruntime import ORTModelForCausalLM

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/gpt2")
>>> model = ORTModelForCausalLM.from_pretrained("optimum/gpt2")
>>> onnx_gen = pipeline("text-generation", model=model, tokenizer=tokenizer)

>>> text = "My name is Arthur and I live in"
>>> gen = onnx_gen(text)

ORTModelForMaskedLM

class optimum.onnxruntime.ORTModelForMaskedLM

< >

( model: InferenceSession config: PretrainedConfig use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None **kwargs )

ONNX Model with a MaskedLMOutput for masked language modeling tasks. This class officially supports albert, bert, camembert, convbert, data2vec_text, deberta, deberta_v2, distilbert, electra, flaubert, ibert, mobilebert, roberta, roformer, squeezebert, xlm, xlm_roberta.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( input_ids: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None attention_mask: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None token_type_ids: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None **kwargs )

Parameters

  • input_ids (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Indices of input sequence tokens in the vocabulary. Indices can be obtained using AutoTokenizer. See PreTrainedTokenizer.encode and PreTrainedTokenizer.__call__ for details. What are input IDs?
  • attention_mask (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:
  • token_type_ids (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:

The ORTModelForMaskedLM forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of feature extraction:

>>> from transformers import AutoTokenizer
>>> from optimum.onnxruntime import ORTModelForMaskedLM
>>> import torch

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/bert-base-uncased-for-fill-mask")
>>> model = ORTModelForMaskedLM.from_pretrained("optimum/bert-base-uncased-for-fill-mask")

>>> inputs = tokenizer("The capital of France is [MASK].", return_tensors="np")

>>> outputs = model(**inputs)
>>> logits = outputs.logits
>>> list(logits.shape)
[1, 8, 28996]

Example using transformers.pipeline:

>>> from transformers import AutoTokenizer, pipeline
>>> from optimum.onnxruntime import ORTModelForMaskedLM

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/bert-base-uncased-for-fill-mask")
>>> model = ORTModelForMaskedLM.from_pretrained("optimum/bert-base-uncased-for-fill-mask")
>>> fill_masker = pipeline("fill-mask", model=model, tokenizer=tokenizer)

>>> text = "The capital of France is [MASK]."
>>> pred = fill_masker(text)

ORTModelForSeq2SeqLM

class optimum.onnxruntime.ORTModelForSeq2SeqLM

< >

( encoder_session: InferenceSession decoder_session: InferenceSession config: PretrainedConfig onnx_paths: typing.List[str] decoder_with_past_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None use_cache: bool = True use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None generation_config: typing.Optional[transformers.generation.configuration_utils.GenerationConfig] = None **kwargs )

Sequence-to-sequence model with a language modeling head for ONNX Runtime inference. This class officially supports bart, blenderbot, blenderbot_small, longt5, m2m_100, marian, mbart, mt5, pegasus, t5.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( input_ids: LongTensor = None attention_mask: typing.Optional[torch.FloatTensor] = None decoder_input_ids: typing.Optional[torch.LongTensor] = None encoder_outputs: typing.Optional[typing.Tuple[typing.Tuple[torch.Tensor]]] = None past_key_values: typing.Optional[typing.Tuple[typing.Tuple[torch.Tensor]]] = None labels: typing.Optional[torch.LongTensor] = None **kwargs )

Parameters

  • input_ids (torch.LongTensor) — Indices of input sequence tokens in the vocabulary of shape (batch_size, encoder_sequence_length).
  • attention_mask (torch.LongTensor) — Mask to avoid performing attention on padding token indices, of shape (batch_size, encoder_sequence_length). Mask values selected in [0, 1].
  • decoder_input_ids (torch.LongTensor) — Indices of decoder input sequence tokens in the vocabulary of shape (batch_size, decoder_sequence_length).
  • encoder_outputs (torch.FloatTensor) — The encoder last_hidden_state of shape (batch_size, encoder_sequence_length, hidden_size).
  • past_key_values (tuple(tuple(torch.FloatTensor), *optional*, defaults to None) — Contains the precomputed key and value hidden states of the attention blocks used to speed up decoding. The tuple is of length config.n_layers with each tuple having 2 tensors of shape (batch_size, num_heads, decoder_sequence_length, embed_size_per_head) and 2 additional tensors of shape (batch_size, num_heads, encoder_sequence_length, embed_size_per_head).

The ORTModelForSeq2SeqLM forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of text generation:

>>> from transformers import AutoTokenizer
>>> from optimum.onnxruntime import ORTModelForSeq2SeqLM

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/t5-small")
>>> model = ORTModelForSeq2SeqLM.from_pretrained("optimum/t5-small")

>>> inputs = tokenizer("My name is Eustache and I like to", return_tensors="pt")

>>> gen_tokens = model.generate(**inputs)
>>> outputs = tokenizer.batch_decode(gen_tokens)

Example using transformers.pipeline:

>>> from transformers import AutoTokenizer, pipeline
>>> from optimum.onnxruntime import ORTModelForSeq2SeqLM

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/t5-small")
>>> model = ORTModelForSeq2SeqLM.from_pretrained("optimum/t5-small")
>>> onnx_translation = pipeline("translation_en_to_de", model=model, tokenizer=tokenizer)

>>> text = "My name is Eustache."
>>> pred = onnx_translation(text)

ORTModelForSequenceClassification

class optimum.onnxruntime.ORTModelForSequenceClassification

< >

( model: InferenceSession config: PretrainedConfig use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None **kwargs )

ONNX Model with a sequence classification/regression head on top (a linear layer on top of the pooled output) e.g. for GLUE tasks. This class officially supports albert, bart, bert, camembert, convbert, data2vec_text, deberta, deberta_v2, distilbert, electra, flaubert, ibert, mbart, mobilebert, nystromformer, roberta, roformer, squeezebert, xlm, xlm_roberta.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( input_ids: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None attention_mask: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None token_type_ids: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None **kwargs )

Parameters

  • input_ids (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Indices of input sequence tokens in the vocabulary. Indices can be obtained using AutoTokenizer. See PreTrainedTokenizer.encode and PreTrainedTokenizer.__call__ for details. What are input IDs?
  • attention_mask (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:
  • token_type_ids (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:

The ORTModelForSequenceClassification forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of single-label classification:

>>> from transformers import AutoTokenizer
>>> from optimum.onnxruntime import ORTModelForSequenceClassification
>>> import torch

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/distilbert-base-uncased-finetuned-sst-2-english")
>>> model = ORTModelForSequenceClassification.from_pretrained("optimum/distilbert-base-uncased-finetuned-sst-2-english")

>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="np")

>>> outputs = model(**inputs)
>>> logits = outputs.logits
>>> list(logits.shape)
[1, 2]

Example using transformers.pipelines:

>>> from transformers import AutoTokenizer, pipeline
>>> from optimum.onnxruntime import ORTModelForSequenceClassification

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/distilbert-base-uncased-finetuned-sst-2-english")
>>> model = ORTModelForSequenceClassification.from_pretrained("optimum/distilbert-base-uncased-finetuned-sst-2-english")
>>> onnx_classifier = pipeline("text-classification", model=model, tokenizer=tokenizer)

>>> text = "Hello, my dog is cute"
>>> pred = onnx_classifier(text)

Example using zero-shot-classification transformers.pipelines:

>>> from transformers import AutoTokenizer, pipeline
>>> from optimum.onnxruntime import ORTModelForSequenceClassification

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/distilbert-base-uncased-mnli")
>>> model = ORTModelForSequenceClassification.from_pretrained("optimum/distilbert-base-uncased-mnli")
>>> onnx_z0 = pipeline("zero-shot-classification", model=model, tokenizer=tokenizer)

>>> sequence_to_classify = "Who are you voting for in 2020?"
>>> candidate_labels = ["Europe", "public health", "politics", "elections"]
>>> pred = onnx_z0(sequence_to_classify, candidate_labels, multi_label=True)

ORTModelForTokenClassification

class optimum.onnxruntime.ORTModelForTokenClassification

< >

( model: InferenceSession config: PretrainedConfig use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None **kwargs )

ONNX 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. This class officially supports albert, bert, bloom, camembert, convbert, data2vec_text, deberta, deberta_v2, distilbert, electra, flaubert, gpt2, ibert, mobilebert, roberta, roformer, squeezebert, xlm, xlm_roberta.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( input_ids: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None attention_mask: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None token_type_ids: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None **kwargs )

Parameters

  • input_ids (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Indices of input sequence tokens in the vocabulary. Indices can be obtained using AutoTokenizer. See PreTrainedTokenizer.encode and PreTrainedTokenizer.__call__ for details. What are input IDs?
  • attention_mask (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:
  • token_type_ids (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:

The ORTModelForTokenClassification forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of token classification:

>>> from transformers import AutoTokenizer
>>> from optimum.onnxruntime import ORTModelForTokenClassification
>>> import torch

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/bert-base-NER")
>>> model = ORTModelForTokenClassification.from_pretrained("optimum/bert-base-NER")

>>> inputs = tokenizer("My name is Philipp and I live in Germany.", return_tensors="np")

>>> outputs = model(**inputs)
>>> logits = outputs.logits
>>> list(logits.shape)
[1, 12, 9]

Example using transformers.pipelines:

>>> from transformers import AutoTokenizer, pipeline
>>> from optimum.onnxruntime import ORTModelForTokenClassification

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/bert-base-NER")
>>> model = ORTModelForTokenClassification.from_pretrained("optimum/bert-base-NER")
>>> onnx_ner = pipeline("token-classification", model=model, tokenizer=tokenizer)

>>> text = "My name is Philipp and I live in Germany."
>>> pred = onnx_ner(text)

ORTModelForMultipleChoice

class optimum.onnxruntime.ORTModelForMultipleChoice

< >

( model: InferenceSession config: PretrainedConfig use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None **kwargs )

ONNX Model with a multiple choice classification head on top (a linear layer on top of the pooled output and a softmax) e.g. for RocStories/SWAG tasks. This class officially supports albert, bert, camembert, convbert, data2vec_text, deberta_v2, distilbert, electra, flaubert, ibert, mobilebert, nystromformer, roberta, roformer, squeezebert, xlm, xlm_roberta.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( input_ids: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None attention_mask: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None token_type_ids: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None **kwargs )

Parameters

  • input_ids (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Indices of input sequence tokens in the vocabulary. Indices can be obtained using AutoTokenizer. See PreTrainedTokenizer.encode and PreTrainedTokenizer.__call__ for details. What are input IDs?
  • attention_mask (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:
  • token_type_ids (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:

The ORTModelForMultipleChoice forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of mutliple choice:

>>> from transformers import AutoTokenizer
>>> from optimum.onnxruntime import ORTModelForMultipleChoice

>>> tokenizer = AutoTokenizer.from_pretrained("ehdwns1516/bert-base-uncased_SWAG")
>>> model = ORTModelForMultipleChoice.from_pretrained("ehdwns1516/bert-base-uncased_SWAG", export=True)

>>> num_choices = 4
>>> first_sentence = ["Members of the procession walk down the street holding small horn brass instruments."] * num_choices
>>> second_sentence = [
...     "A drum line passes by walking down the street playing their instruments.",
...     "A drum line has heard approaching them.",
...     "A drum line arrives and they're outside dancing and asleep.",
...     "A drum line turns the lead singer watches the performance."
... ]
>>> inputs = tokenizer(first_sentence, second_sentence, truncation=True, padding=True)

# Unflatten the inputs values expanding it to the shape [batch_size, num_choices, seq_length]
>>> for k, v in inputs.items():
...     inputs[k] = [v[i: i + num_choices] for i in range(0, len(v), num_choices)]
>>> inputs = dict(inputs.convert_to_tensors(tensor_type="pt"))
>>> outputs = model(**inputs)
>>> logits = outputs.logits

ORTModelForQuestionAnswering

class optimum.onnxruntime.ORTModelForQuestionAnswering

< >

( model: InferenceSession config: PretrainedConfig use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None **kwargs )

ONNX Model with a QuestionAnsweringModelOutput for extractive question-answering tasks like SQuAD. This class officially supports albert, bart, bert, camembert, convbert, data2vec_text, deberta, deberta_v2, distilbert, electra, flaubert, gptj, ibert, mbart, mobilebert, nystromformer, roberta, roformer, squeezebert, xlm, xlm_roberta.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( input_ids: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None attention_mask: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None token_type_ids: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None **kwargs )

Parameters

  • input_ids (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Indices of input sequence tokens in the vocabulary. Indices can be obtained using AutoTokenizer. See PreTrainedTokenizer.encode and PreTrainedTokenizer.__call__ for details. What are input IDs?
  • attention_mask (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:
  • token_type_ids (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:

The ORTModelForQuestionAnswering forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of question answering:

>>> from transformers import AutoTokenizer
>>> from optimum.onnxruntime import ORTModelForQuestionAnswering
>>> import torch

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/roberta-base-squad2")
>>> model = ORTModelForQuestionAnswering.from_pretrained("optimum/roberta-base-squad2")

>>> question, text = "Who was Jim Henson?", "Jim Henson was a nice puppet"
>>> inputs = tokenizer(question, text, return_tensors="np")
>>> start_positions = torch.tensor([1])
>>> end_positions = torch.tensor([3])

>>> outputs = model(**inputs, start_positions=start_positions, end_positions=end_positions)
>>> start_scores = outputs.start_logits
>>> end_scores = outputs.end_logits

Example using transformers.pipeline:

>>> from transformers import AutoTokenizer, pipeline
>>> from optimum.onnxruntime import ORTModelForQuestionAnswering

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/roberta-base-squad2")
>>> model = ORTModelForQuestionAnswering.from_pretrained("optimum/roberta-base-squad2")
>>> onnx_qa = pipeline("question-answering", model=model, tokenizer=tokenizer)

>>> question, text = "Who was Jim Henson?", "Jim Henson was a nice puppet"
>>> pred = onnx_qa(question, text)

Computer vision

The following ORT classes are available for the following computer vision tasks.

ORTModelForImageClassification

class optimum.onnxruntime.ORTModelForImageClassification

< >

( model: InferenceSession config: PretrainedConfig use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None **kwargs )

ONNX Model for image-classification tasks. This class officially supports beit, convnext, convnextv2, data2vec_vision, deit, levit, mobilenet_v1, mobilenet_v2, mobilevit, poolformer, resnet, segformer, swin, vit.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( pixel_values: typing.Union[torch.Tensor, numpy.ndarray] **kwargs )

Parameters

  • pixel_values (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, num_channels, height, width), defaults to None) — Pixel values corresponding to the images in the current batch. Pixel values can be obtained from encoded images using AutoFeatureExtractor.

The ORTModelForImageClassification forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of image classification:

>>> import requests
>>> from PIL import Image
>>> from optimum.onnxruntime import ORTModelForImageClassification
>>> from transformers import AutoFeatureExtractor

>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> image = Image.open(requests.get(url, stream=True).raw)

>>> preprocessor = AutoFeatureExtractor.from_pretrained("optimum/vit-base-patch16-224")
>>> model = ORTModelForImageClassification.from_pretrained("optimum/vit-base-patch16-224")

>>> inputs = preprocessor(images=image, return_tensors="np")

>>> outputs = model(**inputs)
>>> logits = outputs.logits

Example using transformers.pipeline:

>>> import requests
>>> from PIL import Image
>>> from transformers import AutoFeatureExtractor, pipeline
>>> from optimum.onnxruntime import ORTModelForImageClassification

>>> preprocessor = AutoFeatureExtractor.from_pretrained("optimum/vit-base-patch16-224")
>>> model = ORTModelForImageClassification.from_pretrained("optimum/vit-base-patch16-224")
>>> onnx_image_classifier = pipeline("image-classification", model=model, feature_extractor=preprocessor)

>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> pred = onnx_image_classifier(url)

ORTModelForSemanticSegmentation

class optimum.onnxruntime.ORTModelForSemanticSegmentation

< >

( model: InferenceSession config: PretrainedConfig use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None **kwargs )

ONNX Model for semantic-segmentation, with an all-MLP decode head on top e.g. for ADE20k, CityScapes. This class officially supports segformer.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( pixel_values: typing.Union[torch.Tensor, numpy.ndarray] **kwargs )

Parameters

  • pixel_values (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, num_channels, height, width), defaults to None) — Pixel values corresponding to the images in the current batch. Pixel values can be obtained from encoded images using AutoFeatureExtractor.

The ORTModelForSemanticSegmentation forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of semantic segmentation:

>>> import requests
>>> from PIL import Image
>>> from optimum.onnxruntime import ORTModelForSemanticSegmentation
>>> from transformers import AutoFeatureExtractor

>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> image = Image.open(requests.get(url, stream=True).raw)

>>> preprocessor = AutoFeatureExtractor.from_pretrained("optimum/segformer-b0-finetuned-ade-512-512")
>>> model = ORTModelForSemanticSegmentation.from_pretrained("optimum/segformer-b0-finetuned-ade-512-512")

>>> inputs = preprocessor(images=image, return_tensors="np")

>>> outputs = model(**inputs)
>>> logits = outputs.logits

Example using transformers.pipeline:

>>> import requests
>>> from PIL import Image
>>> from transformers import AutoFeatureExtractor, pipeline
>>> from optimum.onnxruntime import ORTModelForSemanticSegmentation

>>> preprocessor = AutoFeatureExtractor.from_pretrained("optimum/segformer-b0-finetuned-ade-512-512")
>>> model = ORTModelForSemanticSegmentation.from_pretrained("optimum/segformer-b0-finetuned-ade-512-512")
>>> onnx_image_segmenter = pipeline("image-segmentation", model=model, feature_extractor=preprocessor)

>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> pred = onnx_image_segmenter(url)

Audio

The following ORT classes are available for the following audio tasks.

ORTModelForAudioClassification

class optimum.onnxruntime.ORTModelForAudioClassification

< >

( model: InferenceSession config: PretrainedConfig use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None **kwargs )

ONNX Model for audio-classification, with a sequence classification head on top (a linear layer over the pooled output) for tasks like SUPERB Keyword Spotting. This class officially supports audio_spectrogram_transformer, data2vec_audio, hubert, sew, sew_d, unispeech, unispeech_sat, wavlm, wav2vec2, wav2vec2-conformer.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( input_values: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None attention_mask: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None input_features: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None **kwargs )

Parameters

  • input_values (torch.Tensor of shape (batch_size, sequence_length)) — Float values of input raw speech waveform.. Input values can be obtained from audio file loaded into an array using AutoFeatureExtractor.

The ORTModelForAudioClassification forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of audio classification:

>>> from transformers import AutoFeatureExtractor
>>> from optimum.onnxruntime import ORTModelForAudioClassification
>>> from datasets import load_dataset
>>> import torch

>>> dataset = load_dataset("hf-internal-testing/librispeech_asr_demo", "clean", split="validation")
>>> dataset = dataset.sort("id")
>>> sampling_rate = dataset.features["audio"].sampling_rate

>>> feature_extractor = AutoFeatureExtractor.from_pretrained("optimum/hubert-base-superb-ks")
>>> model = ORTModelForAudioClassification.from_pretrained("optimum/hubert-base-superb-ks")

>>> # audio file is decoded on the fly
>>> inputs = feature_extractor(dataset[0]["audio"]["array"], sampling_rate=sampling_rate, return_tensors="pt")

>>> with torch.no_grad():
...     logits = model(**inputs).logits

>>> predicted_class_ids = torch.argmax(logits, dim=-1).item()
>>> predicted_label = model.config.id2label[predicted_class_ids]

Example using transformers.pipeline:

>>> from transformers import AutoFeatureExtractor, pipeline
>>> from optimum.onnxruntime import ORTModelForAudioClassification

>>> feature_extractor = AutoFeatureExtractor.from_pretrained("optimum/hubert-base-superb-ks")
>>> dataset = load_dataset("hf-internal-testing/librispeech_asr_demo", "clean", split="validation")
>>> dataset = dataset.sort("id")

>>> model = ORTModelForAudioClassification.from_pretrained("optimum/hubert-base-superb-ks")
>>> onnx_ac = pipeline("audio-classification", model=model, feature_extractor=feature_extractor)

>>> pred = onnx_ac(dataset[0]["audio"]["array"])

ORTModelForAudioFrameClassification

class optimum.onnxruntime.ORTModelForAudioFrameClassification

< >

( model: InferenceSession config: PretrainedConfig use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None **kwargs )

ONNX Model with a frame classification head on top for tasks like Speaker Diarization. This class officially supports data2vec_audio, unispeech_sat, wavlm, wav2vec2, wav2vec2-conformer.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( input_values: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None **kwargs )

Parameters

  • input_values (torch.Tensor of shape (batch_size, sequence_length)) — Float values of input raw speech waveform.. Input values can be obtained from audio file loaded into an array using AutoFeatureExtractor.

The ORTModelForAudioFrameClassification forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of audio frame classification:

>>> from transformers import AutoFeatureExtractor
>>> from optimum.onnxruntime import ORTModelForAudioFrameClassification
>>> from datasets import load_dataset
>>> import torch

>>> dataset = load_dataset("hf-internal-testing/librispeech_asr_demo", "clean", split="validation")
>>> dataset = dataset.sort("id")
>>> sampling_rate = dataset.features["audio"].sampling_rate

>>> feature_extractor = AutoFeatureExtractor.from_pretrained("optimum/wav2vec2-base-superb-sd")
>>> model =  ORTModelForAudioFrameClassification.from_pretrained("optimum/wav2vec2-base-superb-sd")

>>> inputs = feature_extractor(dataset[0]["audio"]["array"], return_tensors="pt", sampling_rate=sampling_rate)
>>> with torch.no_grad():
...     logits = model(**inputs).logits

>>> probabilities = torch.sigmoid(logits[0])
>>> labels = (probabilities > 0.5).long()
>>> labels[0].tolist()

ORTModelForCTC

class optimum.onnxruntime.ORTModelForCTC

< >

( model: InferenceSession config: PretrainedConfig use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None **kwargs )

ONNX Model with a language modeling head on top for Connectionist Temporal Classification (CTC). This class officially supports data2vec_audio, hubert, sew, sew_d, unispeech, unispeech_sat, wavlm, wav2vec2, wav2vec2-conformer.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( input_values: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None **kwargs )

Parameters

  • input_values (torch.Tensor of shape (batch_size, sequence_length)) — Float values of input raw speech waveform.. Input values can be obtained from audio file loaded into an array using AutoFeatureExtractor.

The ORTModelForCTC forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of CTC:

>>> from transformers import AutoProcessor, HubertForCTC
>>> from optimum.onnxruntime import ORTModelForCTC
>>> from datasets import load_dataset
>>> import torch

>>> dataset = load_dataset("hf-internal-testing/librispeech_asr_demo", "clean", split="validation")
>>> dataset = dataset.sort("id")
>>> sampling_rate = dataset.features["audio"].sampling_rate

>>> processor = AutoProcessor.from_pretrained("optimum/hubert-large-ls960-ft")
>>> model = ORTModelForCTC.from_pretrained("optimum/hubert-large-ls960-ft")

>>> # audio file is decoded on the fly
>>> inputs = processor(dataset[0]["audio"]["array"], sampling_rate=sampling_rate, return_tensors="pt")
>>> with torch.no_grad():
...     logits = model(**inputs).logits
>>> predicted_ids = torch.argmax(logits, dim=-1)

>>> transcription = processor.batch_decode(predicted_ids)

ORTModelForSpeechSeq2Seq

class optimum.onnxruntime.ORTModelForSpeechSeq2Seq

< >

( encoder_session: InferenceSession decoder_session: InferenceSession config: PretrainedConfig onnx_paths: typing.List[str] decoder_with_past_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None use_cache: bool = True use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None generation_config: typing.Optional[transformers.generation.configuration_utils.GenerationConfig] = None **kwargs )

Speech Sequence-to-sequence model with a language modeling head for ONNX Runtime inference. This class officially supports whisper, speech_to_text.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( input_features: typing.Optional[torch.FloatTensor] = None attention_mask: typing.Optional[torch.LongTensor] = None decoder_input_ids: typing.Optional[torch.LongTensor] = None encoder_outputs: typing.Optional[typing.Tuple[typing.Tuple[torch.Tensor]]] = None past_key_values: typing.Optional[typing.Tuple[typing.Tuple[torch.Tensor]]] = None labels: typing.Optional[torch.LongTensor] = None cache_position: typing.Optional[torch.Tensor] = None **kwargs )

Parameters

  • input_features (torch.FloatTensor) — Mel features extracted from the raw speech waveform. (batch_size, feature_size, encoder_sequence_length).
  • decoder_input_ids (torch.LongTensor) — Indices of decoder input sequence tokens in the vocabulary of shape (batch_size, decoder_sequence_length).
  • encoder_outputs (torch.FloatTensor) — The encoder last_hidden_state of shape (batch_size, encoder_sequence_length, hidden_size).
  • past_key_values (tuple(tuple(torch.FloatTensor), *optional*, defaults to None) — Contains the precomputed key and value hidden states of the attention blocks used to speed up decoding. The tuple is of length config.n_layers with each tuple having 2 tensors of shape (batch_size, num_heads, decoder_sequence_length, embed_size_per_head) and 2 additional tensors of shape (batch_size, num_heads, encoder_sequence_length, embed_size_per_head).

The ORTModelForSpeechSeq2Seq forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of text generation:

>>> from transformers import AutoProcessor
>>> from optimum.onnxruntime import ORTModelForSpeechSeq2Seq
>>> from datasets import load_dataset

>>> processor = AutoProcessor.from_pretrained("optimum/whisper-tiny.en")
>>> model = ORTModelForSpeechSeq2Seq.from_pretrained("optimum/whisper-tiny.en")

>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> inputs = processor.feature_extractor(ds[0]["audio"]["array"], return_tensors="pt")

>>> gen_tokens = model.generate(inputs=inputs.input_features)
>>> outputs = processor.tokenizer.batch_decode(gen_tokens)

Example using transformers.pipeline:

>>> from transformers import AutoProcessor, pipeline
>>> from optimum.onnxruntime import ORTModelForSpeechSeq2Seq
>>> from datasets import load_dataset

>>> processor = AutoProcessor.from_pretrained("optimum/whisper-tiny.en")
>>> model = ORTModelForSpeechSeq2Seq.from_pretrained("optimum/whisper-tiny.en")
>>> speech_recognition = pipeline("automatic-speech-recognition", model=model, tokenizer=processor.tokenizer, feature_extractor=processor.feature_extractor)

>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> pred = speech_recognition(ds[0]["audio"]["array"])

ORTModelForAudioXVector

class optimum.onnxruntime.ORTModelForAudioXVector

< >

( model: InferenceSession config: PretrainedConfig use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None **kwargs )

ONNX Model with an XVector feature extraction head on top for tasks like Speaker Verification. This class officially supports data2vec_audio, unispeech_sat, wavlm, wav2vec2, wav2vec2-conformer.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( input_values: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None **kwargs )

Parameters

  • input_values (torch.Tensor of shape (batch_size, sequence_length)) — Float values of input raw speech waveform.. Input values can be obtained from audio file loaded into an array using AutoFeatureExtractor.

The ORTModelForAudioXVector forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of Audio XVector:

>>> from transformers import AutoFeatureExtractor
>>> from optimum.onnxruntime import ORTModelForAudioXVector
>>> from datasets import load_dataset
>>> import torch

>>> dataset = load_dataset("hf-internal-testing/librispeech_asr_demo", "clean", split="validation")
>>> dataset = dataset.sort("id")
>>> sampling_rate = dataset.features["audio"].sampling_rate

>>> feature_extractor = AutoFeatureExtractor.from_pretrained("optimum/wav2vec2-base-superb-sv")
>>> model = ORTModelForAudioXVector.from_pretrained("optimum/wav2vec2-base-superb-sv")

>>> # audio file is decoded on the fly
>>> inputs = feature_extractor(
...     [d["array"] for d in dataset[:2]["audio"]], sampling_rate=sampling_rate, return_tensors="pt", padding=True
... )
>>> with torch.no_grad():
...     embeddings = model(**inputs).embeddings

>>> embeddings = torch.nn.functional.normalize(embeddings, dim=-1).cpu()

>>> cosine_sim = torch.nn.CosineSimilarity(dim=-1)
>>> similarity = cosine_sim(embeddings[0], embeddings[1])
>>> threshold = 0.7
>>> if similarity < threshold:
...     print("Speakers are not the same!")
>>> round(similarity.item(), 2)

Multimodal

The following ORT classes are available for the following multimodal tasks.

ORTModelForVision2Seq

class optimum.onnxruntime.ORTModelForVision2Seq

< >

( encoder_session: InferenceSession decoder_session: InferenceSession config: PretrainedConfig onnx_paths: typing.List[str] decoder_with_past_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None use_cache: bool = True use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None generation_config: typing.Optional[transformers.generation.configuration_utils.GenerationConfig] = None **kwargs )

VisionEncoderDecoder Sequence-to-sequence model with a language modeling head for ONNX Runtime inference. This class officially supports trocr and vision-encoder-decoder.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( pixel_values: typing.Optional[torch.FloatTensor] = None decoder_input_ids: typing.Optional[torch.LongTensor] = None encoder_outputs: typing.Optional[typing.Tuple[typing.Tuple[torch.Tensor]]] = None past_key_values: typing.Optional[typing.Tuple[typing.Tuple[torch.Tensor]]] = None labels: typing.Optional[torch.LongTensor] = None **kwargs )

Parameters

  • pixel_values (torch.FloatTensor) — Features extracted from an Image. This tensor should be of shape (batch_size, num_channels, height, width).
  • decoder_input_ids (torch.LongTensor) — Indices of decoder input sequence tokens in the vocabulary of shape (batch_size, decoder_sequence_length).
  • encoder_outputs (torch.FloatTensor) — The encoder last_hidden_state of shape (batch_size, encoder_sequence_length, hidden_size).
  • past_key_values (tuple(tuple(torch.FloatTensor), *optional*, defaults to None) — Contains the precomputed key and value hidden states of the attention blocks used to speed up decoding. The tuple is of length config.n_layers with each tuple having 2 tensors of shape (batch_size, num_heads, decoder_sequence_length, embed_size_per_head) and 2 additional tensors of shape (batch_size, num_heads, encoder_sequence_length, embed_size_per_head).

The ORTModelForVision2Seq forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of text generation:

>>> from transformers import AutoImageProcessor, AutoTokenizer
>>> from optimum.onnxruntime import ORTModelForVision2Seq
>>> from PIL import Image
>>> import requests


>>> processor = AutoImageProcessor.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
>>> tokenizer = AutoTokenizer.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
>>> model = ORTModelForVision2Seq.from_pretrained("nlpconnect/vit-gpt2-image-captioning", export=True)

>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> image = Image.open(requests.get(url, stream=True).raw)
>>> inputs = processor(image, return_tensors="pt")

>>> gen_tokens = model.generate(**inputs)
>>> outputs = tokenizer.batch_decode(gen_tokens, skip_special_tokens=True)

Example using transformers.pipeline:

>>> from transformers import AutoImageProcessor, AutoTokenizer, pipeline
>>> from optimum.onnxruntime import ORTModelForVision2Seq
>>> from PIL import Image
>>> import requests


>>> processor = AutoImageProcessor.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
>>> tokenizer = AutoTokenizer.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
>>> model = ORTModelForVision2Seq.from_pretrained("nlpconnect/vit-gpt2-image-captioning", export=True)

>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> image = Image.open(requests.get(url, stream=True).raw)

>>> image_to_text = pipeline("image-to-text", model=model, tokenizer=tokenizer, feature_extractor=processor, image_processor=processor)
>>> pred = image_to_text(image)

ORTModelForPix2Struct

class optimum.onnxruntime.ORTModelForPix2Struct

< >

( encoder_session: InferenceSession decoder_session: InferenceSession config: PretrainedConfig onnx_paths: typing.List[str] decoder_with_past_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None use_cache: bool = True use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None generation_config: typing.Optional[transformers.generation.configuration_utils.GenerationConfig] = None **kwargs )

Pix2struct model with a language modeling head for ONNX Runtime inference. This class officially supports pix2struct.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( flattened_patches: typing.Optional[torch.FloatTensor] = None attention_mask: typing.Optional[torch.LongTensor] = None decoder_input_ids: typing.Optional[torch.LongTensor] = None decoder_attention_mask: typing.Optional[torch.BoolTensor] = None encoder_outputs: typing.Optional[typing.Tuple[typing.Tuple[torch.Tensor]]] = None past_key_values: typing.Optional[typing.Tuple[typing.Tuple[torch.Tensor]]] = None labels: typing.Optional[torch.LongTensor] = None **kwargs )

Parameters

  • flattened_patches (torch.FloatTensor of shape (batch_size, seq_length, hidden_size)) — Flattened pixel patches. the hidden_size is obtained by the following formula: hidden_size = num_channels patch_size patch_size The process of flattening the pixel patches is done by Pix2StructProcessor.
  • attention_mask (torch.FloatTensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices.
  • decoder_input_ids (torch.LongTensor of shape (batch_size, target_sequence_length), optional) — Indices of decoder input sequence tokens in the vocabulary. Pix2StructText uses the pad_token_id as the starting token for decoder_input_ids generation. If past_key_values is used, optionally only the last decoder_input_ids have to be input (see past_key_values).
  • decoder_attention_mask (torch.BoolTensor of shape (batch_size, target_sequence_length), optional) — Default behavior: generate a tensor that ignores pad tokens in decoder_input_ids. Causal mask will also be used by default.
  • encoder_outputs (tuple(tuple(torch.FloatTensor), optional) — Tuple consists of (last_hidden_state, optional: hidden_states, optional: attentions) last_hidden_state of shape (batch_size, sequence_length, hidden_size) is a sequence of hidden states at the output of the last layer of the encoder. Used in the cross-attention of the decoder.
  • past_key_values (tuple(tuple(torch.FloatTensor), *optional*, defaults to None) — Contains the precomputed key and value hidden states of the attention blocks used to speed up decoding. The tuple is of length config.n_layers with each tuple having 2 tensors of shape (batch_size, num_heads, decoder_sequence_length, embed_size_per_head) and 2 additional tensors of shape (batch_size, num_heads, encoder_sequence_length, embed_size_per_head).

The ORTModelForPix2Struct forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of pix2struct:

>>> from transformers import AutoProcessor
>>> from optimum.onnxruntime import ORTModelForPix2Struct
>>> from PIL import Image
>>> import requests

>>> processor = AutoProcessor.from_pretrained("google/pix2struct-ai2d-base")
>>> model = ORTModelForPix2Struct.from_pretrained("google/pix2struct-ai2d-base", export=True, use_io_binding=True)

>>> url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/ai2d-demo.jpg"
>>> image = Image.open(requests.get(url, stream=True).raw)
>>> question = "What does the label 15 represent? (1) lava (2) core (3) tunnel (4) ash cloud"
>>> inputs = processor(images=image, text=question, return_tensors="pt")

>>> gen_tokens = model.generate(**inputs)
>>> outputs = processor.batch_decode(gen_tokens, skip_special_tokens=True)

Custom Tasks

The following ORT classes are available for the following custom tasks.

ORTModelForCustomTasks

class optimum.onnxruntime.ORTModelForCustomTasks

< >

( model: InferenceSession config: PretrainedConfig use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None **kwargs )

ONNX Model for any custom tasks. It can be used to leverage the inference acceleration for any single-file ONNX model, that may use custom inputs and outputs.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( **model_inputs: typing.Union[torch.Tensor, numpy.ndarray] )

The ORTModelForCustomTasks forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of custom tasks(e.g. a sentence transformers taking pooler_output as output):

>>> from transformers import AutoTokenizer
>>> from optimum.onnxruntime import ORTModelForCustomTasks

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/sbert-all-MiniLM-L6-with-pooler")
>>> model = ORTModelForCustomTasks.from_pretrained("optimum/sbert-all-MiniLM-L6-with-pooler")

>>> inputs = tokenizer("I love burritos!", return_tensors="np")

>>> outputs = model(**inputs)
>>> last_hidden_state = outputs.last_hidden_state
>>> pooler_output = outputs.pooler_output

Example using transformers.pipelines(only if the task is supported):

>>> from transformers import AutoTokenizer, pipeline
>>> from optimum.onnxruntime import ORTModelForCustomTasks

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/sbert-all-MiniLM-L6-with-pooler")
>>> model = ORTModelForCustomTasks.from_pretrained("optimum/sbert-all-MiniLM-L6-with-pooler")
>>> onnx_extractor = pipeline("feature-extraction", model=model, tokenizer=tokenizer)

>>> text = "I love burritos!"
>>> pred = onnx_extractor(text)

ORTModelForFeatureExtraction

class optimum.onnxruntime.ORTModelForFeatureExtraction

< >

( model: InferenceSession config: PretrainedConfig use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None preprocessors: typing.Optional[typing.List] = None **kwargs )

ONNX Model for feature-extraction task.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

forward

< >

( input_ids: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None attention_mask: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None token_type_ids: typing.Union[torch.Tensor, numpy.ndarray, NoneType] = None **kwargs )

Parameters

  • input_ids (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Indices of input sequence tokens in the vocabulary. Indices can be obtained using AutoTokenizer. See PreTrainedTokenizer.encode and PreTrainedTokenizer.__call__ for details. What are input IDs?
  • attention_mask (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:
  • token_type_ids (Union[torch.Tensor, np.ndarray, None] of shape (batch_size, sequence_length), defaults to None) — Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:

The ORTModelForFeatureExtraction forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example of feature extraction:

>>> from transformers import AutoTokenizer
>>> from optimum.onnxruntime import ORTModelForFeatureExtraction
>>> import torch

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/all-MiniLM-L6-v2")
>>> model = ORTModelForFeatureExtraction.from_pretrained("optimum/all-MiniLM-L6-v2")

>>> inputs = tokenizer("My name is Philipp and I live in Germany.", return_tensors="np")

>>> outputs = model(**inputs)
>>> last_hidden_state = outputs.last_hidden_state
>>> list(last_hidden_state.shape)
[1, 12, 384]

Example using transformers.pipeline:

>>> from transformers import AutoTokenizer, pipeline
>>> from optimum.onnxruntime import ORTModelForFeatureExtraction

>>> tokenizer = AutoTokenizer.from_pretrained("optimum/all-MiniLM-L6-v2")
>>> model = ORTModelForFeatureExtraction.from_pretrained("optimum/all-MiniLM-L6-v2")
>>> onnx_extractor = pipeline("feature-extraction", model=model, tokenizer=tokenizer)

>>> text = "My name is Philipp and I live in Germany."
>>> pred = onnx_extractor(text)

Stable Diffusion

ORTStableDiffusionPipeline

class optimum.onnxruntime.ORTStableDiffusionPipeline

< >

( config: typing.Dict[str, typing.Any] tokenizer: CLIPTokenizer scheduler: SchedulerMixin unet_session: InferenceSession feature_extractor: typing.Optional[transformers.models.clip.feature_extraction_clip.CLIPFeatureExtractor] = None vae_encoder_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None vae_decoder_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None text_encoder_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None text_encoder_2_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None tokenizer_2: typing.Optional[transformers.models.clip.tokenization_clip.CLIPTokenizer] = None use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None **kwargs )

ONNX Runtime-powered stable diffusion pipeline corresponding to diffusers.StableDiffusionPipeline.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

__call__

< >

( prompt: typing.Union[str, typing.List[str], NoneType] = None height: typing.Optional[int] = None width: typing.Optional[int] = None num_inference_steps: int = 50 timesteps: typing.List[int] = None sigmas: typing.List[float] = None guidance_scale: float = 7.5 negative_prompt: typing.Union[str, typing.List[str], NoneType] = None num_images_per_prompt: typing.Optional[int] = 1 eta: float = 0.0 generator: typing.Union[torch._C.Generator, typing.List[torch._C.Generator], NoneType] = None latents: typing.Optional[torch.Tensor] = None prompt_embeds: typing.Optional[torch.Tensor] = None negative_prompt_embeds: typing.Optional[torch.Tensor] = None ip_adapter_image: typing.Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, typing.List[PIL.Image.Image], typing.List[numpy.ndarray], typing.List[torch.Tensor], NoneType] = None ip_adapter_image_embeds: typing.Optional[typing.List[torch.Tensor]] = None output_type: typing.Optional[str] = 'pil' return_dict: bool = True cross_attention_kwargs: typing.Union[typing.Dict[str, typing.Any], NoneType] = None guidance_rescale: float = 0.0 clip_skip: typing.Optional[int] = None callback_on_step_end: typing.Union[typing.Callable[[int, int, typing.Dict], NoneType], diffusers.callbacks.PipelineCallback, diffusers.callbacks.MultiPipelineCallbacks, NoneType] = None callback_on_step_end_tensor_inputs: typing.List[str] = ['latents'] **kwargs ) ~pipelines.stable_diffusion.StableDiffusionPipelineOutput or tuple

Parameters

  • prompt (str or List[str], optional) — The prompt or prompts to guide image generation. If not defined, you need to pass prompt_embeds.
  • height (int, optional, defaults to self.unet.config.sample_size * self.vae_scale_factor) — The height in pixels of the generated image.
  • width (int, optional, defaults to self.unet.config.sample_size * self.vae_scale_factor) — The width in pixels of the generated image.
  • num_inference_steps (int, optional, defaults to 50) — The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference.
  • timesteps (List[int], optional) — Custom timesteps to use for the denoising process with schedulers which support a timesteps argument in their set_timesteps method. If not defined, the default behavior when num_inference_steps is passed will be used. Must be in descending order.
  • sigmas (List[float], optional) — Custom sigmas to use for the denoising process with schedulers which support a sigmas argument in their set_timesteps method. If not defined, the default behavior when num_inference_steps is passed will be used.
  • guidance_scale (float, optional, defaults to 7.5) — A higher guidance scale value encourages the model to generate images closely linked to the text prompt at the expense of lower image quality. Guidance scale is enabled when guidance_scale > 1.
  • negative_prompt (str or List[str], optional) — The prompt or prompts to guide what to not include in image generation. If not defined, you need to pass negative_prompt_embeds instead. Ignored when not using guidance (guidance_scale < 1).
  • num_images_per_prompt (int, optional, defaults to 1) — The number of images to generate per prompt.
  • eta (float, optional, defaults to 0.0) — Corresponds to parameter eta (η) from the DDIM paper. Only applies to the ~schedulers.DDIMScheduler, and is ignored in other schedulers.
  • generator (torch.Generator or List[torch.Generator], optional) — A torch.Generator to make generation deterministic.
  • latents (torch.Tensor, optional) — Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image generation. Can be used to tweak the same generation with different prompts. If not provided, a latents tensor is generated by sampling using the supplied random generator.
  • prompt_embeds (torch.Tensor, optional) — Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not provided, text embeddings are generated from the prompt input argument.
  • negative_prompt_embeds (torch.Tensor, optional) — Pre-generated negative text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not provided, negative_prompt_embeds are generated from the negative_prompt input argument. ip_adapter_image — (PipelineImageInput, optional): Optional image input to work with IP Adapters.
  • ip_adapter_image_embeds (List[torch.Tensor], optional) — Pre-generated image embeddings for IP-Adapter. It should be a list of length same as number of IP-adapters. Each element should be a tensor of shape (batch_size, num_images, emb_dim). It should contain the negative image embedding if do_classifier_free_guidance is set to True. If not provided, embeddings are computed from the ip_adapter_image input argument.
  • output_type (str, optional, defaults to "pil") — The output format of the generated image. Choose between PIL.Image or np.array.
  • return_dict (bool, optional, defaults to True) — Whether or not to return a ~pipelines.stable_diffusion.StableDiffusionPipelineOutput instead of a plain tuple.
  • cross_attention_kwargs (dict, optional) — A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined in self.processor.
  • guidance_rescale (float, optional, defaults to 0.0) — Guidance rescale factor from Common Diffusion Noise Schedules and Sample Steps are Flawed. Guidance rescale factor should fix overexposure when using zero terminal SNR.
  • clip_skip (int, optional) — Number of layers to be skipped from CLIP while computing the prompt embeddings. A value of 1 means that the output of the pre-final layer will be used for computing the prompt embeddings.
  • callback_on_step_end (Callable, PipelineCallback, MultiPipelineCallbacks, optional) — A function or a subclass of PipelineCallback or MultiPipelineCallbacks that is called at the end of each denoising step during the inference. with the following arguments: callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict). callback_kwargs will include a list of all tensors as specified by callback_on_step_end_tensor_inputs.
  • callback_on_step_end_tensor_inputs (List, optional) — The list of tensor inputs for the callback_on_step_end function. The tensors specified in the list will be passed as callback_kwargs argument. You will only be able to include variables listed in the ._callback_tensor_inputs attribute of your pipeline class.

Returns

~pipelines.stable_diffusion.StableDiffusionPipelineOutput or tuple

If return_dict is True, ~pipelines.stable_diffusion.StableDiffusionPipelineOutput is returned, otherwise a tuple is returned where the first element is a list with the generated images and the second element is a list of bools indicating whether the corresponding generated image contains “not-safe-for-work” (nsfw) content.

The call function to the pipeline for generation.

ORTStableDiffusionImg2ImgPipeline

class optimum.onnxruntime.ORTStableDiffusionImg2ImgPipeline

< >

( config: typing.Dict[str, typing.Any] tokenizer: CLIPTokenizer scheduler: SchedulerMixin unet_session: InferenceSession feature_extractor: typing.Optional[transformers.models.clip.feature_extraction_clip.CLIPFeatureExtractor] = None vae_encoder_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None vae_decoder_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None text_encoder_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None text_encoder_2_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None tokenizer_2: typing.Optional[transformers.models.clip.tokenization_clip.CLIPTokenizer] = None use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None **kwargs )

ONNX Runtime-powered stable diffusion pipeline corresponding to diffusers.StableDiffusionImg2ImgPipeline.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

__call__

< >

( prompt: typing.Union[str, typing.List[str]] = None image: typing.Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, typing.List[PIL.Image.Image], typing.List[numpy.ndarray], typing.List[torch.Tensor]] = None strength: float = 0.8 num_inference_steps: typing.Optional[int] = 50 timesteps: typing.List[int] = None sigmas: typing.List[float] = None guidance_scale: typing.Optional[float] = 7.5 negative_prompt: typing.Union[str, typing.List[str], NoneType] = None num_images_per_prompt: typing.Optional[int] = 1 eta: typing.Optional[float] = 0.0 generator: typing.Union[torch._C.Generator, typing.List[torch._C.Generator], NoneType] = None prompt_embeds: typing.Optional[torch.Tensor] = None negative_prompt_embeds: typing.Optional[torch.Tensor] = None ip_adapter_image: typing.Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, typing.List[PIL.Image.Image], typing.List[numpy.ndarray], typing.List[torch.Tensor], NoneType] = None ip_adapter_image_embeds: typing.Optional[typing.List[torch.Tensor]] = None output_type: typing.Optional[str] = 'pil' return_dict: bool = True cross_attention_kwargs: typing.Union[typing.Dict[str, typing.Any], NoneType] = None clip_skip: int = None callback_on_step_end: typing.Union[typing.Callable[[int, int, typing.Dict], NoneType], diffusers.callbacks.PipelineCallback, diffusers.callbacks.MultiPipelineCallbacks, NoneType] = None callback_on_step_end_tensor_inputs: typing.List[str] = ['latents'] **kwargs ) ~pipelines.stable_diffusion.StableDiffusionPipelineOutput or tuple

Parameters

  • prompt (str or List[str], optional) — The prompt or prompts to guide image generation. If not defined, you need to pass prompt_embeds.
  • image (torch.Tensor, PIL.Image.Image, np.ndarray, List[torch.Tensor], List[PIL.Image.Image], or List[np.ndarray]) — Image, numpy array or tensor representing an image batch to be used as the starting point. For both numpy array and pytorch tensor, the expected value range is between [0, 1] If it’s a tensor or a list or tensors, the expected shape should be (B, C, H, W) or (C, H, W). If it is a numpy array or a list of arrays, the expected shape should be (B, H, W, C) or (H, W, C) It can also accept image latents as image, but if passing latents directly it is not encoded again.
  • strength (float, optional, defaults to 0.8) — Indicates extent to transform the reference image. Must be between 0 and 1. image is used as a starting point and more noise is added the higher the strength. The number of denoising steps depends on the amount of noise initially added. When strength is 1, added noise is maximum and the denoising process runs for the full number of iterations specified in num_inference_steps. A value of 1 essentially ignores image.
  • num_inference_steps (int, optional, defaults to 50) — The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference. This parameter is modulated by strength.
  • timesteps (List[int], optional) — Custom timesteps to use for the denoising process with schedulers which support a timesteps argument in their set_timesteps method. If not defined, the default behavior when num_inference_steps is passed will be used. Must be in descending order.
  • sigmas (List[float], optional) — Custom sigmas to use for the denoising process with schedulers which support a sigmas argument in their set_timesteps method. If not defined, the default behavior when num_inference_steps is passed will be used.
  • guidance_scale (float, optional, defaults to 7.5) — A higher guidance scale value encourages the model to generate images closely linked to the text prompt at the expense of lower image quality. Guidance scale is enabled when guidance_scale > 1.
  • negative_prompt (str or List[str], optional) — The prompt or prompts to guide what to not include in image generation. If not defined, you need to pass negative_prompt_embeds instead. Ignored when not using guidance (guidance_scale < 1).
  • num_images_per_prompt (int, optional, defaults to 1) — The number of images to generate per prompt.
  • eta (float, optional, defaults to 0.0) — Corresponds to parameter eta (η) from the DDIM paper. Only applies to the ~schedulers.DDIMScheduler, and is ignored in other schedulers.
  • generator (torch.Generator or List[torch.Generator], optional) — A torch.Generator to make generation deterministic.
  • prompt_embeds (torch.Tensor, optional) — Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not provided, text embeddings are generated from the prompt input argument.
  • negative_prompt_embeds (torch.Tensor, optional) — Pre-generated negative text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not provided, negative_prompt_embeds are generated from the negative_prompt input argument. ip_adapter_image — (PipelineImageInput, optional): Optional image input to work with IP Adapters.
  • ip_adapter_image_embeds (List[torch.Tensor], optional) — Pre-generated image embeddings for IP-Adapter. It should be a list of length same as number of IP-adapters. Each element should be a tensor of shape (batch_size, num_images, emb_dim). It should contain the negative image embedding if do_classifier_free_guidance is set to True. If not provided, embeddings are computed from the ip_adapter_image input argument.
  • output_type (str, optional, defaults to "pil") — The output format of the generated image. Choose between PIL.Image or np.array.
  • return_dict (bool, optional, defaults to True) — Whether or not to return a ~pipelines.stable_diffusion.StableDiffusionPipelineOutput instead of a plain tuple.
  • cross_attention_kwargs (dict, optional) — A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined in self.processor.
  • clip_skip (int, optional) — Number of layers to be skipped from CLIP while computing the prompt embeddings. A value of 1 means that the output of the pre-final layer will be used for computing the prompt embeddings.
  • callback_on_step_end (Callable, PipelineCallback, MultiPipelineCallbacks, optional) — A function or a subclass of PipelineCallback or MultiPipelineCallbacks that is called at the end of each denoising step during the inference. with the following arguments: callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict). callback_kwargs will include a list of all tensors as specified by callback_on_step_end_tensor_inputs.
  • callback_on_step_end_tensor_inputs (List, optional) — The list of tensor inputs for the callback_on_step_end function. The tensors specified in the list will be passed as callback_kwargs argument. You will only be able to include variables listed in the ._callback_tensor_inputs attribute of your pipeline class.

Returns

~pipelines.stable_diffusion.StableDiffusionPipelineOutput or tuple

If return_dict is True, ~pipelines.stable_diffusion.StableDiffusionPipelineOutput is returned, otherwise a tuple is returned where the first element is a list with the generated images and the second element is a list of bools indicating whether the corresponding generated image contains “not-safe-for-work” (nsfw) content.

The call function to the pipeline for generation.

Examples:

ORTStableDiffusionInpaintPipeline

class optimum.onnxruntime.ORTStableDiffusionInpaintPipeline

< >

( config: typing.Dict[str, typing.Any] tokenizer: CLIPTokenizer scheduler: SchedulerMixin unet_session: InferenceSession feature_extractor: typing.Optional[transformers.models.clip.feature_extraction_clip.CLIPFeatureExtractor] = None vae_encoder_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None vae_decoder_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None text_encoder_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None text_encoder_2_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None tokenizer_2: typing.Optional[transformers.models.clip.tokenization_clip.CLIPTokenizer] = None use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None **kwargs )

ONNX Runtime-powered stable diffusion pipeline corresponding to diffusers.StableDiffusionInpaintPipeline.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

__call__

< >

( prompt: typing.Union[str, typing.List[str]] = None image: typing.Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, typing.List[PIL.Image.Image], typing.List[numpy.ndarray], typing.List[torch.Tensor]] = None mask_image: typing.Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, typing.List[PIL.Image.Image], typing.List[numpy.ndarray], typing.List[torch.Tensor]] = None masked_image_latents: torch.Tensor = None height: typing.Optional[int] = None width: typing.Optional[int] = None padding_mask_crop: typing.Optional[int] = None strength: float = 1.0 num_inference_steps: int = 50 timesteps: typing.List[int] = None sigmas: typing.List[float] = None guidance_scale: float = 7.5 negative_prompt: typing.Union[str, typing.List[str], NoneType] = None num_images_per_prompt: typing.Optional[int] = 1 eta: float = 0.0 generator: typing.Union[torch._C.Generator, typing.List[torch._C.Generator], NoneType] = None latents: typing.Optional[ForwardRef('torch.Tensor')] = None prompt_embeds: typing.Optional[ForwardRef('torch.Tensor')] = None negative_prompt_embeds: typing.Optional[ForwardRef('torch.Tensor')] = None ip_adapter_image: typing.Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, typing.List[PIL.Image.Image], typing.List[numpy.ndarray], typing.List[torch.Tensor], NoneType] = None ip_adapter_image_embeds: typing.Optional[typing.List[ForwardRef('torch.Tensor')]] = None output_type: typing.Optional[str] = 'pil' return_dict: bool = True cross_attention_kwargs: typing.Union[typing.Dict[str, typing.Any], NoneType] = None clip_skip: int = None callback_on_step_end: typing.Union[typing.Callable[[int, int, typing.Dict], NoneType], diffusers.callbacks.PipelineCallback, diffusers.callbacks.MultiPipelineCallbacks, NoneType] = None callback_on_step_end_tensor_inputs: typing.List[str] = ['latents'] **kwargs ) ~pipelines.stable_diffusion.StableDiffusionPipelineOutput or tuple

Parameters

  • prompt (str or List[str], optional) — The prompt or prompts to guide image generation. If not defined, you need to pass prompt_embeds.
  • image (torch.Tensor, PIL.Image.Image, np.ndarray, List[torch.Tensor], List[PIL.Image.Image], or List[np.ndarray]) — Image, numpy array or tensor representing an image batch to be inpainted (which parts of the image to be masked out with mask_image and repainted according to prompt). For both numpy array and pytorch tensor, the expected value range is between [0, 1] If it’s a tensor or a list or tensors, the expected shape should be (B, C, H, W) or (C, H, W). If it is a numpy array or a list of arrays, the expected shape should be (B, H, W, C) or (H, W, C) It can also accept image latents as image, but if passing latents directly it is not encoded again.
  • mask_image (torch.Tensor, PIL.Image.Image, np.ndarray, List[torch.Tensor], List[PIL.Image.Image], or List[np.ndarray]) — Image, numpy array or tensor representing an image batch to mask image. White pixels in the mask are repainted while black pixels are preserved. If mask_image is a PIL image, it is converted to a single channel (luminance) before use. If it’s a numpy array or pytorch tensor, it should contain one color channel (L) instead of 3, so the expected shape for pytorch tensor would be (B, 1, H, W), (B, H, W), (1, H, W), (H, W). And for numpy array would be for (B, H, W, 1), (B, H, W), (H, W, 1), or (H, W).
  • height (int, optional, defaults to self.unet.config.sample_size * self.vae_scale_factor) — The height in pixels of the generated image.
  • width (int, optional, defaults to self.unet.config.sample_size * self.vae_scale_factor) — The width in pixels of the generated image.
  • padding_mask_crop (int, optional, defaults to None) — The size of margin in the crop to be applied to the image and masking. If None, no crop is applied to image and mask_image. If padding_mask_crop is not None, it will first find a rectangular region with the same aspect ration of the image and contains all masked area, and then expand that area based on padding_mask_crop. The image and mask_image will then be cropped based on the expanded area before resizing to the original image size for inpainting. This is useful when the masked area is small while the image is large and contain information irrelevant for inpainting, such as background.
  • strength (float, optional, defaults to 1.0) — Indicates extent to transform the reference image. Must be between 0 and 1. image is used as a starting point and more noise is added the higher the strength. The number of denoising steps depends on the amount of noise initially added. When strength is 1, added noise is maximum and the denoising process runs for the full number of iterations specified in num_inference_steps. A value of 1 essentially ignores image.
  • num_inference_steps (int, optional, defaults to 50) — The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference. This parameter is modulated by strength.
  • timesteps (List[int], optional) — Custom timesteps to use for the denoising process with schedulers which support a timesteps argument in their set_timesteps method. If not defined, the default behavior when num_inference_steps is passed will be used. Must be in descending order.
  • sigmas (List[float], optional) — Custom sigmas to use for the denoising process with schedulers which support a sigmas argument in their set_timesteps method. If not defined, the default behavior when num_inference_steps is passed will be used.
  • guidance_scale (float, optional, defaults to 7.5) — A higher guidance scale value encourages the model to generate images closely linked to the text prompt at the expense of lower image quality. Guidance scale is enabled when guidance_scale > 1.
  • negative_prompt (str or List[str], optional) — The prompt or prompts to guide what to not include in image generation. If not defined, you need to pass negative_prompt_embeds instead. Ignored when not using guidance (guidance_scale < 1).
  • num_images_per_prompt (int, optional, defaults to 1) — The number of images to generate per prompt.
  • eta (float, optional, defaults to 0.0) — Corresponds to parameter eta (η) from the DDIM paper. Only applies to the ~schedulers.DDIMScheduler, and is ignored in other schedulers.
  • generator (torch.Generator or List[torch.Generator], optional) — A torch.Generator to make generation deterministic.
  • latents (torch.Tensor, optional) — Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image generation. Can be used to tweak the same generation with different prompts. If not provided, a latents tensor is generated by sampling using the supplied random generator.
  • prompt_embeds (torch.Tensor, optional) — Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not provided, text embeddings are generated from the prompt input argument.
  • negative_prompt_embeds (torch.Tensor, optional) — Pre-generated negative text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not provided, negative_prompt_embeds are generated from the negative_prompt input argument. ip_adapter_image — (PipelineImageInput, optional): Optional image input to work with IP Adapters.
  • ip_adapter_image_embeds (List[torch.Tensor], optional) — Pre-generated image embeddings for IP-Adapter. It should be a list of length same as number of IP-adapters. Each element should be a tensor of shape (batch_size, num_images, emb_dim). It should contain the negative image embedding if do_classifier_free_guidance is set to True. If not provided, embeddings are computed from the ip_adapter_image input argument.
  • output_type (str, optional, defaults to "pil") — The output format of the generated image. Choose between PIL.Image or np.array.
  • return_dict (bool, optional, defaults to True) — Whether or not to return a ~pipelines.stable_diffusion.StableDiffusionPipelineOutput instead of a plain tuple.
  • cross_attention_kwargs (dict, optional) — A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined in self.processor.
  • clip_skip (int, optional) — Number of layers to be skipped from CLIP while computing the prompt embeddings. A value of 1 means that the output of the pre-final layer will be used for computing the prompt embeddings.
  • callback_on_step_end (Callable, PipelineCallback, MultiPipelineCallbacks, optional) — A function or a subclass of PipelineCallback or MultiPipelineCallbacks that is called at the end of each denoising step during the inference. with the following arguments: callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict). callback_kwargs will include a list of all tensors as specified by callback_on_step_end_tensor_inputs.
  • callback_on_step_end_tensor_inputs (List, optional) — The list of tensor inputs for the callback_on_step_end function. The tensors specified in the list will be passed as callback_kwargs argument. You will only be able to include variables listed in the ._callback_tensor_inputs attribute of your pipeline class.

Returns

~pipelines.stable_diffusion.StableDiffusionPipelineOutput or tuple

If return_dict is True, ~pipelines.stable_diffusion.StableDiffusionPipelineOutput is returned, otherwise a tuple is returned where the first element is a list with the generated images and the second element is a list of bools indicating whether the corresponding generated image contains “not-safe-for-work” (nsfw) content.

The call function to the pipeline for generation.

Examples:

>>> import PIL
>>> import requests
>>> import torch
>>> from io import BytesIO

>>> from diffusers import StableDiffusionInpaintPipeline


>>> def download_image(url):
...     response = requests.get(url)
...     return PIL.Image.open(BytesIO(response.content)).convert("RGB")


>>> img_url = "https://raw.githubusercontent.com/CompVis/latent-diffusion/main/data/inpainting_examples/overture-creations-5sI6fQgYIuo.png"
>>> mask_url = "https://raw.githubusercontent.com/CompVis/latent-diffusion/main/data/inpainting_examples/overture-creations-5sI6fQgYIuo_mask.png"

>>> init_image = download_image(img_url).resize((512, 512))
>>> mask_image = download_image(mask_url).resize((512, 512))

>>> pipe = StableDiffusionInpaintPipeline.from_pretrained(
...     "runwayml/stable-diffusion-inpainting", torch_dtype=torch.float16
... )
>>> pipe = pipe.to("cuda")

>>> prompt = "Face of a yellow cat, high resolution, sitting on a park bench"
>>> image = pipe(prompt=prompt, image=init_image, mask_image=mask_image).images[0]

ORTStableDiffusionXLPipeline

class optimum.onnxruntime.ORTStableDiffusionXLPipeline

< >

( *args add_watermarker: typing.Optional[bool] = None **kwargs )

ONNX Runtime-powered stable diffusion pipeline corresponding to diffusers.StableDiffusionXLPipeline.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

__call__

< >

( prompt: typing.Union[str, typing.List[str]] = None prompt_2: typing.Union[str, typing.List[str], NoneType] = None height: typing.Optional[int] = None width: typing.Optional[int] = None num_inference_steps: int = 50 timesteps: typing.List[int] = None sigmas: typing.List[float] = None denoising_end: typing.Optional[float] = None guidance_scale: float = 5.0 negative_prompt: typing.Union[str, typing.List[str], NoneType] = None negative_prompt_2: typing.Union[str, typing.List[str], NoneType] = None num_images_per_prompt: typing.Optional[int] = 1 eta: float = 0.0 generator: typing.Union[torch._C.Generator, typing.List[torch._C.Generator], NoneType] = None latents: typing.Optional[torch.Tensor] = None prompt_embeds: typing.Optional[torch.Tensor] = None negative_prompt_embeds: typing.Optional[torch.Tensor] = None pooled_prompt_embeds: typing.Optional[torch.Tensor] = None negative_pooled_prompt_embeds: typing.Optional[torch.Tensor] = None ip_adapter_image: typing.Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, typing.List[PIL.Image.Image], typing.List[numpy.ndarray], typing.List[torch.Tensor], NoneType] = None ip_adapter_image_embeds: typing.Optional[typing.List[torch.Tensor]] = None output_type: typing.Optional[str] = 'pil' return_dict: bool = True cross_attention_kwargs: typing.Union[typing.Dict[str, typing.Any], NoneType] = None guidance_rescale: float = 0.0 original_size: typing.Union[typing.Tuple[int, int], NoneType] = None crops_coords_top_left: typing.Tuple[int, int] = (0, 0) target_size: typing.Union[typing.Tuple[int, int], NoneType] = None negative_original_size: typing.Union[typing.Tuple[int, int], NoneType] = None negative_crops_coords_top_left: typing.Tuple[int, int] = (0, 0) negative_target_size: typing.Union[typing.Tuple[int, int], NoneType] = None clip_skip: typing.Optional[int] = None callback_on_step_end: typing.Union[typing.Callable[[int, int, typing.Dict], NoneType], diffusers.callbacks.PipelineCallback, diffusers.callbacks.MultiPipelineCallbacks, NoneType] = None callback_on_step_end_tensor_inputs: typing.List[str] = ['latents'] **kwargs ) ~pipelines.stable_diffusion_xl.StableDiffusionXLPipelineOutput or tuple

Parameters

  • prompt (str or List[str], optional) — The prompt or prompts to guide the image generation. If not defined, one has to pass prompt_embeds. instead.
  • prompt_2 (str or List[str], optional) — The prompt or prompts to be sent to the tokenizer_2 and text_encoder_2. If not defined, prompt is used in both text-encoders
  • height (int, optional, defaults to self.unet.config.sample_size * self.vae_scale_factor) — The height in pixels of the generated image. This is set to 1024 by default for the best results. Anything below 512 pixels won’t work well for stabilityai/stable-diffusion-xl-base-1.0 and checkpoints that are not specifically fine-tuned on low resolutions.
  • width (int, optional, defaults to self.unet.config.sample_size * self.vae_scale_factor) — The width in pixels of the generated image. This is set to 1024 by default for the best results. Anything below 512 pixels won’t work well for stabilityai/stable-diffusion-xl-base-1.0 and checkpoints that are not specifically fine-tuned on low resolutions.
  • num_inference_steps (int, optional, defaults to 50) — The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference.
  • timesteps (List[int], optional) — Custom timesteps to use for the denoising process with schedulers which support a timesteps argument in their set_timesteps method. If not defined, the default behavior when num_inference_steps is passed will be used. Must be in descending order.
  • sigmas (List[float], optional) — Custom sigmas to use for the denoising process with schedulers which support a sigmas argument in their set_timesteps method. If not defined, the default behavior when num_inference_steps is passed will be used.
  • denoising_end (float, optional) — When specified, determines the fraction (between 0.0 and 1.0) of the total denoising process to be completed before it is intentionally prematurely terminated. As a result, the returned sample will still retain a substantial amount of noise as determined by the discrete timesteps selected by the scheduler. The denoising_end parameter should ideally be utilized when this pipeline forms a part of a “Mixture of Denoisers” multi-pipeline setup, as elaborated in Refining the Image Output
  • guidance_scale (float, optional, defaults to 5.0) — Guidance scale as defined in Classifier-Free Diffusion Guidance. guidance_scale is defined as w of equation 2. of Imagen Paper. Guidance scale is enabled by setting guidance_scale > 1. Higher guidance scale encourages to generate images that are closely linked to the text prompt, usually at the expense of lower image quality.
  • negative_prompt (str or List[str], optional) — The prompt or prompts not to guide the image generation. If not defined, one has to pass negative_prompt_embeds instead. Ignored when not using guidance (i.e., ignored if guidance_scale is less than 1).
  • negative_prompt_2 (str or List[str], optional) — The prompt or prompts not to guide the image generation to be sent to tokenizer_2 and text_encoder_2. If not defined, negative_prompt is used in both text-encoders
  • num_images_per_prompt (int, optional, defaults to 1) — The number of images to generate per prompt.
  • eta (float, optional, defaults to 0.0) — Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to schedulers.DDIMScheduler, will be ignored for others.
  • generator (torch.Generator or List[torch.Generator], optional) — One or a list of torch generator(s) to make generation deterministic.
  • latents (torch.Tensor, optional) — Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image generation. Can be used to tweak the same generation with different prompts. If not provided, a latents tensor will ge generated by sampling using the supplied random generator.
  • prompt_embeds (torch.Tensor, optional) — Pre-generated text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, text embeddings will be generated from prompt input argument.
  • negative_prompt_embeds (torch.Tensor, optional) — Pre-generated negative text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, negative_prompt_embeds will be generated from negative_prompt input argument.
  • pooled_prompt_embeds (torch.Tensor, optional) — Pre-generated pooled text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, pooled text embeddings will be generated from prompt input argument.
  • negative_pooled_prompt_embeds (torch.Tensor, optional) — Pre-generated negative pooled text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, pooled negative_prompt_embeds will be generated from negative_prompt input argument. ip_adapter_image — (PipelineImageInput, optional): Optional image input to work with IP Adapters.
  • ip_adapter_image_embeds (List[torch.Tensor], optional) — Pre-generated image embeddings for IP-Adapter. It should be a list of length same as number of IP-adapters. Each element should be a tensor of shape (batch_size, num_images, emb_dim). It should contain the negative image embedding if do_classifier_free_guidance is set to True. If not provided, embeddings are computed from the ip_adapter_image input argument.
  • output_type (str, optional, defaults to "pil") — The output format of the generate image. Choose between PIL: PIL.Image.Image or np.array.
  • return_dict (bool, optional, defaults to True) — Whether or not to return a ~pipelines.stable_diffusion_xl.StableDiffusionXLPipelineOutput instead of a plain tuple.
  • cross_attention_kwargs (dict, optional) — A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined under self.processor in diffusers.models.attention_processor.
  • guidance_rescale (float, optional, defaults to 0.0) — Guidance rescale factor proposed by Common Diffusion Noise Schedules and Sample Steps are Flawed guidance_scale is defined as φ in equation 16. of Common Diffusion Noise Schedules and Sample Steps are Flawed. Guidance rescale factor should fix overexposure when using zero terminal SNR.
  • original_size (Tuple[int], optional, defaults to (1024, 1024)) — If original_size is not the same as target_size the image will appear to be down- or upsampled. original_size defaults to (height, width) if not specified. Part of SDXL’s micro-conditioning as explained in section 2.2 of https://huggingface.co/papers/2307.01952.
  • crops_coords_top_left (Tuple[int], optional, defaults to (0, 0)) — crops_coords_top_left can be used to generate an image that appears to be “cropped” from the position crops_coords_top_left downwards. Favorable, well-centered images are usually achieved by setting crops_coords_top_left to (0, 0). Part of SDXL’s micro-conditioning as explained in section 2.2 of https://huggingface.co/papers/2307.01952.
  • target_size (Tuple[int], optional, defaults to (1024, 1024)) — For most cases, target_size should be set to the desired height and width of the generated image. If not specified it will default to (height, width). Part of SDXL’s micro-conditioning as explained in section 2.2 of https://huggingface.co/papers/2307.01952.
  • negative_original_size (Tuple[int], optional, defaults to (1024, 1024)) — To negatively condition the generation process based on a specific image resolution. Part of SDXL’s micro-conditioning as explained in section 2.2 of https://huggingface.co/papers/2307.01952. For more information, refer to this issue thread: https://github.com/huggingface/diffusers/issues/4208.
  • negative_crops_coords_top_left (Tuple[int], optional, defaults to (0, 0)) — To negatively condition the generation process based on a specific crop coordinates. Part of SDXL’s micro-conditioning as explained in section 2.2 of https://huggingface.co/papers/2307.01952. For more information, refer to this issue thread: https://github.com/huggingface/diffusers/issues/4208.
  • negative_target_size (Tuple[int], optional, defaults to (1024, 1024)) — To negatively condition the generation process based on a target image resolution. It should be as same as the target_size for most cases. Part of SDXL’s micro-conditioning as explained in section 2.2 of https://huggingface.co/papers/2307.01952. For more information, refer to this issue thread: https://github.com/huggingface/diffusers/issues/4208.
  • callback_on_step_end (Callable, PipelineCallback, MultiPipelineCallbacks, optional) — A function or a subclass of PipelineCallback or MultiPipelineCallbacks that is called at the end of each denoising step during the inference. with the following arguments: callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict). callback_kwargs will include a list of all tensors as specified by callback_on_step_end_tensor_inputs.
  • callback_on_step_end_tensor_inputs (List, optional) — The list of tensor inputs for the callback_on_step_end function. The tensors specified in the list will be passed as callback_kwargs argument. You will only be able to include variables listed in the ._callback_tensor_inputs attribute of your pipeline class.

Returns

~pipelines.stable_diffusion_xl.StableDiffusionXLPipelineOutput or tuple

~pipelines.stable_diffusion_xl.StableDiffusionXLPipelineOutput if return_dict is True, otherwise a tuple. When returning a tuple, the first element is a list with the generated images.

Function invoked when calling the pipeline for generation.

Examples:

ORTStableDiffusionXLImg2ImgPipeline

class optimum.onnxruntime.ORTStableDiffusionXLImg2ImgPipeline

< >

( *args add_watermarker: typing.Optional[bool] = None **kwargs )

ONNX Runtime-powered stable diffusion pipeline corresponding to diffusers.StableDiffusionXLImg2ImgPipeline.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

__call__

< >

( prompt: typing.Union[str, typing.List[str]] = None prompt_2: typing.Union[str, typing.List[str], NoneType] = None image: typing.Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, typing.List[PIL.Image.Image], typing.List[numpy.ndarray], typing.List[torch.Tensor]] = None strength: float = 0.3 num_inference_steps: int = 50 timesteps: typing.List[int] = None sigmas: typing.List[float] = None denoising_start: typing.Optional[float] = None denoising_end: typing.Optional[float] = None guidance_scale: float = 5.0 negative_prompt: typing.Union[str, typing.List[str], NoneType] = None negative_prompt_2: typing.Union[str, typing.List[str], NoneType] = None num_images_per_prompt: typing.Optional[int] = 1 eta: float = 0.0 generator: typing.Union[torch._C.Generator, typing.List[torch._C.Generator], NoneType] = None latents: typing.Optional[torch.Tensor] = None prompt_embeds: typing.Optional[torch.Tensor] = None negative_prompt_embeds: typing.Optional[torch.Tensor] = None pooled_prompt_embeds: typing.Optional[torch.Tensor] = None negative_pooled_prompt_embeds: typing.Optional[torch.Tensor] = None ip_adapter_image: typing.Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, typing.List[PIL.Image.Image], typing.List[numpy.ndarray], typing.List[torch.Tensor], NoneType] = None ip_adapter_image_embeds: typing.Optional[typing.List[torch.Tensor]] = None output_type: typing.Optional[str] = 'pil' return_dict: bool = True cross_attention_kwargs: typing.Union[typing.Dict[str, typing.Any], NoneType] = None guidance_rescale: float = 0.0 original_size: typing.Tuple[int, int] = None crops_coords_top_left: typing.Tuple[int, int] = (0, 0) target_size: typing.Tuple[int, int] = None negative_original_size: typing.Union[typing.Tuple[int, int], NoneType] = None negative_crops_coords_top_left: typing.Tuple[int, int] = (0, 0) negative_target_size: typing.Union[typing.Tuple[int, int], NoneType] = None aesthetic_score: float = 6.0 negative_aesthetic_score: float = 2.5 clip_skip: typing.Optional[int] = None callback_on_step_end: typing.Union[typing.Callable[[int, int, typing.Dict], NoneType], diffusers.callbacks.PipelineCallback, diffusers.callbacks.MultiPipelineCallbacks, NoneType] = None callback_on_step_end_tensor_inputs: typing.List[str] = ['latents'] **kwargs ) ~pipelines.stable_diffusion.StableDiffusionXLPipelineOutput or tuple

Parameters

  • prompt (str or List[str], optional) — The prompt or prompts to guide the image generation. If not defined, one has to pass prompt_embeds. instead.
  • prompt_2 (str or List[str], optional) — The prompt or prompts to be sent to the tokenizer_2 and text_encoder_2. If not defined, prompt is used in both text-encoders
  • image (torch.Tensor or PIL.Image.Image or np.ndarray or List[torch.Tensor] or List[PIL.Image.Image] or List[np.ndarray]) — The image(s) to modify with the pipeline.
  • strength (float, optional, defaults to 0.3) — Conceptually, indicates how much to transform the reference image. Must be between 0 and 1. image will be used as a starting point, adding more noise to it the larger the strength. The number of denoising steps depends on the amount of noise initially added. When strength is 1, added noise will be maximum and the denoising process will run for the full number of iterations specified in num_inference_steps. A value of 1, therefore, essentially ignores image. Note that in the case of denoising_start being declared as an integer, the value of strength will be ignored.
  • num_inference_steps (int, optional, defaults to 50) — The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference.
  • timesteps (List[int], optional) — Custom timesteps to use for the denoising process with schedulers which support a timesteps argument in their set_timesteps method. If not defined, the default behavior when num_inference_steps is passed will be used. Must be in descending order.
  • sigmas (List[float], optional) — Custom sigmas to use for the denoising process with schedulers which support a sigmas argument in their set_timesteps method. If not defined, the default behavior when num_inference_steps is passed will be used.
  • denoising_start (float, optional) — When specified, indicates the fraction (between 0.0 and 1.0) of the total denoising process to be bypassed before it is initiated. Consequently, the initial part of the denoising process is skipped and it is assumed that the passed image is a partly denoised image. Note that when this is specified, strength will be ignored. The denoising_start parameter is particularly beneficial when this pipeline is integrated into a “Mixture of Denoisers” multi-pipeline setup, as detailed in Refine Image Quality.
  • denoising_end (float, optional) — When specified, determines the fraction (between 0.0 and 1.0) of the total denoising process to be completed before it is intentionally prematurely terminated. As a result, the returned sample will still retain a substantial amount of noise (ca. final 20% of timesteps still needed) and should be denoised by a successor pipeline that has denoising_start set to 0.8 so that it only denoises the final 20% of the scheduler. The denoising_end parameter should ideally be utilized when this pipeline forms a part of a “Mixture of Denoisers” multi-pipeline setup, as elaborated in Refine Image Quality.
  • guidance_scale (float, optional, defaults to 7.5) — Guidance scale as defined in Classifier-Free Diffusion Guidance. guidance_scale is defined as w of equation 2. of Imagen Paper. Guidance scale is enabled by setting guidance_scale > 1. Higher guidance scale encourages to generate images that are closely linked to the text prompt, usually at the expense of lower image quality.
  • negative_prompt (str or List[str], optional) — The prompt or prompts not to guide the image generation. If not defined, one has to pass negative_prompt_embeds instead. Ignored when not using guidance (i.e., ignored if guidance_scale is less than 1).
  • negative_prompt_2 (str or List[str], optional) — The prompt or prompts not to guide the image generation to be sent to tokenizer_2 and text_encoder_2. If not defined, negative_prompt is used in both text-encoders
  • num_images_per_prompt (int, optional, defaults to 1) — The number of images to generate per prompt.
  • eta (float, optional, defaults to 0.0) — Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to schedulers.DDIMScheduler, will be ignored for others.
  • generator (torch.Generator or List[torch.Generator], optional) — One or a list of torch generator(s) to make generation deterministic.
  • latents (torch.Tensor, optional) — Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image generation. Can be used to tweak the same generation with different prompts. If not provided, a latents tensor will ge generated by sampling using the supplied random generator.
  • prompt_embeds (torch.Tensor, optional) — Pre-generated text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, text embeddings will be generated from prompt input argument.
  • negative_prompt_embeds (torch.Tensor, optional) — Pre-generated negative text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, negative_prompt_embeds will be generated from negative_prompt input argument.
  • pooled_prompt_embeds (torch.Tensor, optional) — Pre-generated pooled text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, pooled text embeddings will be generated from prompt input argument.
  • negative_pooled_prompt_embeds (torch.Tensor, optional) — Pre-generated negative pooled text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, pooled negative_prompt_embeds will be generated from negative_prompt input argument. ip_adapter_image — (PipelineImageInput, optional): Optional image input to work with IP Adapters.
  • ip_adapter_image_embeds (List[torch.Tensor], optional) — Pre-generated image embeddings for IP-Adapter. It should be a list of length same as number of IP-adapters. Each element should be a tensor of shape (batch_size, num_images, emb_dim). It should contain the negative image embedding if do_classifier_free_guidance is set to True. If not provided, embeddings are computed from the ip_adapter_image input argument.
  • output_type (str, optional, defaults to "pil") — The output format of the generate image. Choose between PIL: PIL.Image.Image or np.array.
  • return_dict (bool, optional, defaults to True) — Whether or not to return a ~pipelines.stable_diffusion.StableDiffusionXLPipelineOutput instead of a plain tuple.
  • cross_attention_kwargs (dict, optional) — A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined under self.processor in diffusers.models.attention_processor.
  • guidance_rescale (float, optional, defaults to 0.0) — Guidance rescale factor proposed by Common Diffusion Noise Schedules and Sample Steps are Flawed guidance_scale is defined as φ in equation 16. of Common Diffusion Noise Schedules and Sample Steps are Flawed. Guidance rescale factor should fix overexposure when using zero terminal SNR.
  • original_size (Tuple[int], optional, defaults to (1024, 1024)) — If original_size is not the same as target_size the image will appear to be down- or upsampled. original_size defaults to (height, width) if not specified. Part of SDXL’s micro-conditioning as explained in section 2.2 of https://huggingface.co/papers/2307.01952.
  • crops_coords_top_left (Tuple[int], optional, defaults to (0, 0)) — crops_coords_top_left can be used to generate an image that appears to be “cropped” from the position crops_coords_top_left downwards. Favorable, well-centered images are usually achieved by setting crops_coords_top_left to (0, 0). Part of SDXL’s micro-conditioning as explained in section 2.2 of https://huggingface.co/papers/2307.01952.
  • target_size (Tuple[int], optional, defaults to (1024, 1024)) — For most cases, target_size should be set to the desired height and width of the generated image. If not specified it will default to (height, width). Part of SDXL’s micro-conditioning as explained in section 2.2 of https://huggingface.co/papers/2307.01952.
  • negative_original_size (Tuple[int], optional, defaults to (1024, 1024)) — To negatively condition the generation process based on a specific image resolution. Part of SDXL’s micro-conditioning as explained in section 2.2 of https://huggingface.co/papers/2307.01952. For more information, refer to this issue thread: https://github.com/huggingface/diffusers/issues/4208.
  • negative_crops_coords_top_left (Tuple[int], optional, defaults to (0, 0)) — To negatively condition the generation process based on a specific crop coordinates. Part of SDXL’s micro-conditioning as explained in section 2.2 of https://huggingface.co/papers/2307.01952. For more information, refer to this issue thread: https://github.com/huggingface/diffusers/issues/4208.
  • negative_target_size (Tuple[int], optional, defaults to (1024, 1024)) — To negatively condition the generation process based on a target image resolution. It should be as same as the target_size for most cases. Part of SDXL’s micro-conditioning as explained in section 2.2 of https://huggingface.co/papers/2307.01952. For more information, refer to this issue thread: https://github.com/huggingface/diffusers/issues/4208.
  • aesthetic_score (float, optional, defaults to 6.0) — Used to simulate an aesthetic score of the generated image by influencing the positive text condition. Part of SDXL’s micro-conditioning as explained in section 2.2 of https://huggingface.co/papers/2307.01952.
  • negative_aesthetic_score (float, optional, defaults to 2.5) — Part of SDXL’s micro-conditioning as explained in section 2.2 of https://huggingface.co/papers/2307.01952. Can be used to simulate an aesthetic score of the generated image by influencing the negative text condition.
  • clip_skip (int, optional) — Number of layers to be skipped from CLIP while computing the prompt embeddings. A value of 1 means that the output of the pre-final layer will be used for computing the prompt embeddings.
  • callback_on_step_end (Callable, PipelineCallback, MultiPipelineCallbacks, optional) — A function or a subclass of PipelineCallback or MultiPipelineCallbacks that is called at the end of each denoising step during the inference. with the following arguments: callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict). callback_kwargs will include a list of all tensors as specified by callback_on_step_end_tensor_inputs.
  • callback_on_step_end_tensor_inputs (List, optional) — The list of tensor inputs for the callback_on_step_end function. The tensors specified in the list will be passed as callback_kwargs argument. You will only be able to include variables listed in the ._callback_tensor_inputs attribute of your pipeline class.

Returns

~pipelines.stable_diffusion.StableDiffusionXLPipelineOutput or tuple

~pipelines.stable_diffusion.StableDiffusionXLPipelineOutput if return_dict is True, otherwise a `tuple. When returning a tuple, the first element is a list with the generated images.

Function invoked when calling the pipeline for generation.

Examples:

ORTLatentConsistencyModelPipeline

class optimum.onnxruntime.ORTLatentConsistencyModelPipeline

< >

( config: typing.Dict[str, typing.Any] tokenizer: CLIPTokenizer scheduler: SchedulerMixin unet_session: InferenceSession feature_extractor: typing.Optional[transformers.models.clip.feature_extraction_clip.CLIPFeatureExtractor] = None vae_encoder_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None vae_decoder_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None text_encoder_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None text_encoder_2_session: typing.Optional[onnxruntime.capi.onnxruntime_inference_collection.InferenceSession] = None tokenizer_2: typing.Optional[transformers.models.clip.tokenization_clip.CLIPTokenizer] = None use_io_binding: typing.Optional[bool] = None model_save_dir: typing.Union[str, pathlib.Path, tempfile.TemporaryDirectory, NoneType] = None **kwargs )

ONNX Runtime-powered stable diffusion pipeline corresponding to diffusers.LatentConsistencyModelPipeline.

This model inherits from ORTModel, check its documentation for the generic methods the library implements for all its model (such as downloading or saving).

This class should be initialized using the onnxruntime.modeling_ort.ORTModel.from_pretrained() method.

__call__

< >

( prompt: typing.Union[str, typing.List[str]] = None height: typing.Optional[int] = None width: typing.Optional[int] = None num_inference_steps: int = 4 original_inference_steps: int = None timesteps: typing.List[int] = None guidance_scale: float = 8.5 num_images_per_prompt: typing.Optional[int] = 1 generator: typing.Union[torch._C.Generator, typing.List[torch._C.Generator], NoneType] = None latents: typing.Optional[torch.Tensor] = None prompt_embeds: typing.Optional[torch.Tensor] = None ip_adapter_image: typing.Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, typing.List[PIL.Image.Image], typing.List[numpy.ndarray], typing.List[torch.Tensor], NoneType] = None ip_adapter_image_embeds: typing.Optional[typing.List[torch.Tensor]] = None output_type: typing.Optional[str] = 'pil' return_dict: bool = True cross_attention_kwargs: typing.Union[typing.Dict[str, typing.Any], NoneType] = None clip_skip: typing.Optional[int] = None callback_on_step_end: typing.Union[typing.Callable[[int, int, typing.Dict], NoneType], NoneType] = None callback_on_step_end_tensor_inputs: typing.List[str] = ['latents'] **kwargs ) ~pipelines.stable_diffusion.StableDiffusionPipelineOutput or tuple

Parameters

  • prompt (str or List[str], optional) — The prompt or prompts to guide image generation. If not defined, you need to pass prompt_embeds.
  • height (int, optional, defaults to self.unet.config.sample_size * self.vae_scale_factor) — The height in pixels of the generated image.
  • width (int, optional, defaults to self.unet.config.sample_size * self.vae_scale_factor) — The width in pixels of the generated image.
  • num_inference_steps (int, optional, defaults to 50) — The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference.
  • original_inference_steps (int, optional) — The original number of inference steps use to generate a linearly-spaced timestep schedule, from which we will draw num_inference_steps evenly spaced timesteps from as our final timestep schedule, following the Skipping-Step method in the paper (see Section 4.3). If not set this will default to the scheduler’s original_inference_steps attribute.
  • timesteps (List[int], optional) — Custom timesteps to use for the denoising process. If not defined, equal spaced num_inference_steps timesteps on the original LCM training/distillation timestep schedule are used. Must be in descending order.
  • guidance_scale (float, optional, defaults to 7.5) — A higher guidance scale value encourages the model to generate images closely linked to the text prompt at the expense of lower image quality. Guidance scale is enabled when guidance_scale > 1. Note that the original latent consistency models paper uses a different CFG formulation where the guidance scales are decreased by 1 (so in the paper formulation CFG is enabled when guidance_scale > 0).
  • num_images_per_prompt (int, optional, defaults to 1) — The number of images to generate per prompt.
  • generator (torch.Generator or List[torch.Generator], optional) — A torch.Generator to make generation deterministic.
  • latents (torch.Tensor, optional) — Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image generation. Can be used to tweak the same generation with different prompts. If not provided, a latents tensor is generated by sampling using the supplied random generator.
  • prompt_embeds (torch.Tensor, optional) — Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not provided, text embeddings are generated from the prompt input argument. ip_adapter_image — (PipelineImageInput, optional): Optional image input to work with IP Adapters.
  • ip_adapter_image_embeds (List[torch.Tensor], optional) — Pre-generated image embeddings for IP-Adapter. It should be a list of length same as number of IP-adapters. Each element should be a tensor of shape (batch_size, num_images, emb_dim). It should contain the negative image embedding if do_classifier_free_guidance is set to True. If not provided, embeddings are computed from the ip_adapter_image input argument.
  • output_type (str, optional, defaults to "pil") — The output format of the generated image. Choose between PIL.Image or np.array.
  • return_dict (bool, optional, defaults to True) — Whether or not to return a ~pipelines.stable_diffusion.StableDiffusionPipelineOutput instead of a plain tuple.
  • cross_attention_kwargs (dict, optional) — A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined in self.processor.
  • clip_skip (int, optional) — Number of layers to be skipped from CLIP while computing the prompt embeddings. A value of 1 means that the output of the pre-final layer will be used for computing the prompt embeddings.
  • callback_on_step_end (Callable, optional) — A function that calls at the end of each denoising steps during the inference. The function is called with the following arguments: callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict). callback_kwargs will include a list of all tensors as specified by callback_on_step_end_tensor_inputs.
  • callback_on_step_end_tensor_inputs (List, optional) — The list of tensor inputs for the callback_on_step_end function. The tensors specified in the list will be passed as callback_kwargs argument. You will only be able to include variables listed in the ._callback_tensor_inputs attribute of your pipeline class.

Returns

~pipelines.stable_diffusion.StableDiffusionPipelineOutput or tuple

If return_dict is True, ~pipelines.stable_diffusion.StableDiffusionPipelineOutput is returned, otherwise a tuple is returned where the first element is a list with the generated images and the second element is a list of bools indicating whether the corresponding generated image contains “not-safe-for-work” (nsfw) content.

The call function to the pipeline for generation.

Examples:

< > Update on GitHub