Autoformer
개요
The Autoformer 모델은 Haixu Wu, Jiehui Xu, Jianmin Wang, Mingsheng Long가 제안한 오토포머: 장기 시계열 예측을 위한 자기상관 분해 트랜스포머 라는 논문에서 소개 되었습니다.
이 모델은 트랜스포머를 심층 분해 아키텍처로 확장하여, 예측 과정에서 추세와 계절성 요소를 점진적으로 분해할 수 있습니다.
해당 논문의 초록입니다:
예측 시간을 연장하는 것은 극한 기상 조기 경보 및 장기 에너지 소비 계획과 같은 실제 응용 프로그램에 중요한 요구 사항입니다. 본 논문은 시계열의 장기 예측 문제를 연구합니다. 기존의 트랜스포머 기반 모델들은 장거리 종속성을 발견하기 위해 다양한 셀프 어텐션 메커니즘을 채택합니다. 그러나 장기 미래의 복잡한 시간적 패턴으로 인해 모델이 신뢰할 수 있는 종속성을 찾기 어렵습니다. 또한, 트랜스포머는 긴 시계열의 효율성을 위해 점별 셀프 어텐션의 희소 버전을 채택해야 하므로 정보 활용의 병목 현상이 발생합니다. 우리는 트랜스포머를 넘어서 자기상관 메커니즘을 갖춘 새로운 분해 아키텍처인 Autoformer를 설계했습니다. 우리는 시계열 분해의 전처리 관행을 깨고 이를 심층 모델의 기본 내부 블록으로 혁신했습니다. 이 설계는 Autoformer에 복잡한 시계열에 대한 점진적 분해 능력을 부여합니다. 또한, 확률 과정 이론에서 영감을 받아 시계열의 주기성을 기반으로 자기상관 메커니즘을 설계했으며, 이는 하위 시계열 수준에서 종속성 발견과 표현 집계를 수행합니다. 자기상관은 효율성과 정확도 면에서 셀프 어텐션를 능가합니다. 장기 예측에서 Autoformer는 에너지, 교통, 경제, 날씨, 질병 등 5가지 실용적 응용 분야를 포괄하는 6개 벤치마크에서 38%의 상대적 개선으로 최첨단 정확도를 달성했습니다.
이 모델은 elisim과 kashif에 의해 기여 되었습니다. 원본 코드는 이곳에서 확인할 수 있습니다.
자료
시작하는 데 도움이 되는 Hugging Face와 community 자료 목록(🌎로 표시됨) 입니다. 여기에 포함될 자료를 제출하고 싶으시다면 PR(Pull Request)를 열어주세요. 리뷰 해드리겠습니다! 자료는 기존 자료를 복제하는 대신 새로운 내용을 담고 있어야 합니다.
- HuggingFace 블로그에서 Autoformer 포스트를 확인하세요: 네, 그렇습니다, 트랜스포머는 시계열 예측에 효과적입니다(+ Autoformer)
AutoformerConfig
class transformers.AutoformerConfig
< source >( prediction_length: Optional = None context_length: Optional = None distribution_output: str = 'student_t' loss: str = 'nll' input_size: int = 1 lags_sequence: List = [1, 2, 3, 4, 5, 6, 7] scaling: bool = True num_time_features: int = 0 num_dynamic_real_features: int = 0 num_static_categorical_features: int = 0 num_static_real_features: int = 0 cardinality: Optional = None embedding_dimension: Optional = None d_model: int = 64 encoder_attention_heads: int = 2 decoder_attention_heads: int = 2 encoder_layers: int = 2 decoder_layers: int = 2 encoder_ffn_dim: int = 32 decoder_ffn_dim: int = 32 activation_function: str = 'gelu' dropout: float = 0.1 encoder_layerdrop: float = 0.1 decoder_layerdrop: float = 0.1 attention_dropout: float = 0.1 activation_dropout: float = 0.1 num_parallel_samples: int = 100 init_std: float = 0.02 use_cache: bool = True is_encoder_decoder = True label_length: int = 10 moving_average: int = 25 autocorrelation_factor: int = 3 **kwargs )
Parameters
- prediction_length (
int) — The prediction length for the decoder. In other words, the prediction horizon of the model. - context_length (
int, optional, defaults toprediction_length) — The context length for the encoder. If unset, the context length will be the same as theprediction_length. - distribution_output (
string, optional, defaults to"student_t") — The distribution emission head for the model. Could be either “student_t”, “normal” or “negative_binomial”. - loss (
string, optional, defaults to"nll") — The loss function for the model corresponding to thedistribution_outputhead. For parametric distributions it is the negative log likelihood (nll) - which currently is the only supported one. - input_size (
int, optional, defaults to 1) — The size of the target variable which by default is 1 for univariate targets. Would be > 1 in case of multivariate targets. - lags_sequence (
list[int], optional, defaults to[1, 2, 3, 4, 5, 6, 7]) — The lags of the input time series as covariates often dictated by the frequency. Default is[1, 2, 3, 4, 5, 6, 7]. - scaling (
bool, optional defaults toTrue) — Whether to scale the input targets. - num_time_features (
int, optional, defaults to 0) — The number of time features in the input time series. - num_dynamic_real_features (
int, optional, defaults to 0) — The number of dynamic real valued features. - num_static_categorical_features (
int, optional, defaults to 0) — The number of static categorical features. - num_static_real_features (
int, optional, defaults to 0) — The number of static real valued features. - cardinality (
list[int], optional) — The cardinality (number of different values) for each of the static categorical features. Should be a list of integers, having the same length asnum_static_categorical_features. Cannot beNoneifnum_static_categorical_featuresis > 0. - embedding_dimension (
list[int], optional) — The dimension of the embedding for each of the static categorical features. Should be a list of integers, having the same length asnum_static_categorical_features. Cannot beNoneifnum_static_categorical_featuresis > 0. - d_model (
int, optional, defaults to 64) — Dimensionality of the transformer layers. - encoder_layers (
int, optional, defaults to 2) — Number of encoder layers. - decoder_layers (
int, optional, defaults to 2) — Number of decoder layers. - encoder_attention_heads (
int, optional, defaults to 2) — Number of attention heads for each attention layer in the Transformer encoder. - decoder_attention_heads (
int, optional, defaults to 2) — Number of attention heads for each attention layer in the Transformer decoder. - encoder_ffn_dim (
int, optional, defaults to 32) — Dimension of the “intermediate” (often named feed-forward) layer in encoder. - decoder_ffn_dim (
int, optional, defaults to 32) — Dimension of the “intermediate” (often named feed-forward) layer in decoder. - activation_function (
strorfunction, optional, defaults to"gelu") — The non-linear activation function (function or string) in the encoder and decoder. If string,"gelu"and"relu"are supported. - dropout (
float, optional, defaults to 0.1) — The dropout probability for all fully connected layers in the encoder, and decoder. - encoder_layerdrop (
float, optional, defaults to 0.1) — The dropout probability for the attention and fully connected layers for each encoder layer. - decoder_layerdrop (
float, optional, defaults to 0.1) — The dropout probability for the attention and fully connected layers for each decoder layer. - attention_dropout (
float, optional, defaults to 0.1) — The dropout probability for the attention probabilities. - activation_dropout (
float, optional, defaults to 0.1) — The dropout probability used between the two layers of the feed-forward networks. - num_parallel_samples (
int, optional, defaults to 100) — The number of samples to generate in parallel for each time step of inference. - init_std (
float, optional, defaults to 0.02) — The standard deviation of the truncated normal weight initialization distribution. - use_cache (
bool, optional, defaults toTrue) — Whether to use the past key/values attentions (if applicable to the model) to speed up decoding. - label_length (
int, optional, defaults to 10) — Start token length of the Autoformer decoder, which is used for direct multi-step prediction (i.e. non-autoregressive generation). - moving_average (
int, optional, defaults to 25) — The window size of the moving average. In practice, it’s the kernel size in AvgPool1d of the Decomposition Layer. - autocorrelation_factor (
int, optional, defaults to 3) — “Attention” (i.e. AutoCorrelation mechanism) factor which is used to find top k autocorrelations delays. It’s recommended in the paper to set it to a number between 1 and 5.
This is the configuration class to store the configuration of an AutoformerModel. It is used to instantiate an Autoformer model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the Autoformer huggingface/autoformer-tourism-monthly architecture.
Configuration objects inherit from PretrainedConfig can be used to control the model outputs. Read the documentation from PretrainedConfig for more information.
>>> from transformers import AutoformerConfig, AutoformerModel
>>> # Initializing a default Autoformer configuration
>>> configuration = AutoformerConfig()
>>> # Randomly initializing a model (with random weights) from the configuration
>>> model = AutoformerModel(configuration)
>>> # Accessing the model configuration
>>> configuration = model.configAutoformerModel
class transformers.AutoformerModel
< source >( config: AutoformerConfig )
Parameters
- config (AutoformerConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.
The bare Autoformer Model outputting raw hidden-states without any specific head on top. This model inherits from PreTrainedModel. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.)
This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.
forward
< source >( past_values: Tensor past_time_features: Tensor past_observed_mask: Tensor static_categorical_features: Optional = None static_real_features: Optional = None future_values: Optional = None future_time_features: Optional = None decoder_attention_mask: Optional = None head_mask: Optional = None decoder_head_mask: Optional = None cross_attn_head_mask: Optional = None encoder_outputs: Optional = None past_key_values: Optional = None output_hidden_states: Optional = None output_attentions: Optional = None use_cache: Optional = None return_dict: Optional = None ) → transformers.models.autoformer.modeling_autoformer.AutoformerModelOutput or tuple(torch.FloatTensor)
Parameters
- past_values (
torch.FloatTensorof shape(batch_size, sequence_length)) — Past values of the time series, that serve as context in order to predict the future. These values may contain lags, i.e. additional values from the past which are added in order to serve as “extra context”. Thepast_valuesis what the Transformer encoder gets as input (with optional additional features, such asstatic_categorical_features,static_real_features,past_time_features).The sequence length here is equal to
context_length+max(config.lags_sequence).Missing values need to be replaced with zeros.
- past_time_features (
torch.FloatTensorof shape(batch_size, sequence_length, num_features), optional) — Optional time features, which the model internally will add topast_values. These could be things like “month of year”, “day of the month”, etc. encoded as vectors (for instance as Fourier features). These could also be so-called “age” features, which basically help the model know “at which point in life” a time-series is. Age features have small values for distant past time steps and increase monotonically the more we approach the current time step.These features serve as the “positional encodings” of the inputs. So contrary to a model like BERT, where the position encodings are learned from scratch internally as parameters of the model, the Time Series Transformer requires to provide additional time features.
The Autoformer only learns additional embeddings for
static_categorical_features. - past_observed_mask (
torch.BoolTensorof shape(batch_size, sequence_length), optional) — Boolean mask to indicate whichpast_valueswere observed and which were missing. Mask values selected in[0, 1]:- 1 for values that are observed,
- 0 for values that are missing (i.e. NaNs that were replaced by zeros).
- static_categorical_features (
torch.LongTensorof shape(batch_size, number of static categorical features), optional) — Optional static categorical features for which the model will learn an embedding, which it will add to the values of the time series.Static categorical features are features which have the same value for all time steps (static over time).
A typical example of a static categorical feature is a time series ID.
- static_real_features (
torch.FloatTensorof shape(batch_size, number of static real features), optional) — Optional static real features which the model will add to the values of the time series.Static real features are features which have the same value for all time steps (static over time).
A typical example of a static real feature is promotion information.
- future_values (
torch.FloatTensorof shape(batch_size, prediction_length)) — Future values of the time series, that serve as labels for the model. Thefuture_valuesis what the Transformer needs to learn to output, given thepast_values.See the demo notebook and code snippets for details.
Missing values need to be replaced with zeros.
- future_time_features (
torch.FloatTensorof shape(batch_size, prediction_length, num_features), optional) — Optional time features, which the model internally will add tofuture_values. These could be things like “month of year”, “day of the month”, etc. encoded as vectors (for instance as Fourier features). These could also be so-called “age” features, which basically help the model know “at which point in life” a time-series is. Age features have small values for distant past time steps and increase monotonically the more we approach the current time step.These features serve as the “positional encodings” of the inputs. So contrary to a model like BERT, where the position encodings are learned from scratch internally as parameters of the model, the Time Series Transformer requires to provide additional features.
The Autoformer only learns additional embeddings for
static_categorical_features. - attention_mask (
torch.Tensorof shape(batch_size, sequence_length), optional) — Mask to avoid performing attention on certain token indices. Mask values selected in[0, 1]:- 1 for tokens that are not masked,
- 0 for tokens that are masked.
- decoder_attention_mask (
torch.LongTensorof shape(batch_size, target_sequence_length), optional) — Mask to avoid performing attention on certain token indices. By default, a causal mask will be used, to make sure the model can only look at previous inputs in order to predict the future. - head_mask (
torch.Tensorof shape(encoder_layers, encoder_attention_heads), optional) — Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in[0, 1]:- 1 indicates the head is not masked,
- 0 indicates the head is masked.
- decoder_head_mask (
torch.Tensorof shape(decoder_layers, decoder_attention_heads), optional) — Mask to nullify selected heads of the attention modules in the decoder. Mask values selected in[0, 1]:- 1 indicates the head is not masked,
- 0 indicates the head is masked.
- cross_attn_head_mask (
torch.Tensorof shape(decoder_layers, decoder_attention_heads), optional) — Mask to nullify selected heads of the cross-attention modules. Mask values selected in[0, 1]:- 1 indicates the head is not masked,
- 0 indicates the head is masked.
- encoder_outputs (
tuple(tuple(torch.FloatTensor), optional) — Tuple consists oflast_hidden_state,hidden_states(optional) andattentions(optional)last_hidden_stateof shape(batch_size, sequence_length, hidden_size)(optional) 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, returned whenuse_cache=Trueis passed or whenconfig.use_cache=True) — Tuple oftuple(torch.FloatTensor)of lengthconfig.n_layers, with each tuple having 2 tensors of shape(batch_size, num_heads, sequence_length, embed_size_per_head)) and 2 additional tensors of shape(batch_size, num_heads, encoder_sequence_length, embed_size_per_head).Contains pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used (see
past_key_valuesinput) to speed up sequential decoding.If
past_key_valuesare used, the user can optionally input only the lastdecoder_input_ids(those that don’t have their past key value states given to this model) of shape(batch_size, 1)instead of alldecoder_input_idsof shape(batch_size, sequence_length). - inputs_embeds (
torch.FloatTensorof shape(batch_size, sequence_length, hidden_size), optional) — Optionally, instead of passinginput_idsyou can choose to directly pass an embedded representation. This is useful if you want more control over how to convertinput_idsindices into associated vectors than the model’s internal embedding lookup matrix. - use_cache (
bool, optional) — If set toTrue,past_key_valueskey value states are returned and can be used to speed up decoding (seepast_key_values). - output_attentions (
bool, optional) — Whether or not to return the attentions tensors of all attention layers. Seeattentionsunder returned tensors for more detail. - output_hidden_states (
bool, optional) — Whether or not to return the hidden states of all layers. Seehidden_statesunder returned tensors for more detail. - return_dict (
bool, optional) — Whether or not to return a ModelOutput instead of a plain tuple.
Returns
transformers.models.autoformer.modeling_autoformer.AutoformerModelOutput or tuple(torch.FloatTensor)
A transformers.models.autoformer.modeling_autoformer.AutoformerModelOutput or a tuple of
torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various
elements depending on the configuration (AutoformerConfig) and inputs.
-
last_hidden_state (
torch.FloatTensorof shape(batch_size, sequence_length, hidden_size)) — Sequence of hidden-states at the output of the last layer of the decoder of the model.If
past_key_valuesis used only the last hidden-state of the sequences of shape(batch_size, 1, hidden_size)is output. -
trend (
torch.FloatTensorof shape(batch_size, sequence_length, hidden_size)) — Trend tensor for each time series. -
past_key_values (
tuple(tuple(torch.FloatTensor)), optional, returned whenuse_cache=Trueis passed or whenconfig.use_cache=True) — Tuple oftuple(torch.FloatTensor)of lengthconfig.n_layers, with each tuple having 2 tensors of shape(batch_size, num_heads, sequence_length, embed_size_per_head)) and 2 additional tensors of shape(batch_size, num_heads, encoder_sequence_length, embed_size_per_head).Contains pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used (see
past_key_valuesinput) to speed up sequential decoding. -
decoder_hidden_states (
tuple(torch.FloatTensor), optional, returned whenoutput_hidden_states=Trueis passed or whenconfig.output_hidden_states=True) — Tuple oftorch.FloatTensor(one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape(batch_size, sequence_length, hidden_size).Hidden-states of the decoder at the output of each layer plus the optional initial embedding outputs.
-
decoder_attentions (
tuple(torch.FloatTensor), optional, returned whenoutput_attentions=Trueis passed or whenconfig.output_attentions=True) — Tuple oftorch.FloatTensor(one for each layer) of shape(batch_size, num_heads, sequence_length, sequence_length).Attentions weights of the decoder, after the attention softmax, used to compute the weighted average in the self-attention heads.
-
cross_attentions (
tuple(torch.FloatTensor), optional, returned whenoutput_attentions=Trueis passed or whenconfig.output_attentions=True) — Tuple oftorch.FloatTensor(one for each layer) of shape(batch_size, num_heads, sequence_length, sequence_length).Attentions weights of the decoder’s cross-attention layer, after the attention softmax, used to compute the weighted average in the cross-attention heads.
-
encoder_last_hidden_state (
torch.FloatTensorof shape(batch_size, sequence_length, hidden_size), optional) — Sequence of hidden-states at the output of the last layer of the encoder of the model. -
encoder_hidden_states (
tuple(torch.FloatTensor), optional, returned whenoutput_hidden_states=Trueis passed or whenconfig.output_hidden_states=True) — Tuple oftorch.FloatTensor(one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape(batch_size, sequence_length, hidden_size).Hidden-states of the encoder at the output of each layer plus the optional initial embedding outputs.
-
encoder_attentions (
tuple(torch.FloatTensor), optional, returned whenoutput_attentions=Trueis passed or whenconfig.output_attentions=True) — Tuple oftorch.FloatTensor(one for each layer) of shape(batch_size, num_heads, sequence_length, sequence_length).Attentions weights of the encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.
-
loc (
torch.FloatTensorof shape(batch_size,)or(batch_size, input_size), optional) — Shift values of each time series’ context window which is used to give the model inputs of the same magnitude and then used to shift back to the original magnitude. -
scale (
torch.FloatTensorof shape(batch_size,)or(batch_size, input_size), optional) — Scaling values of each time series’ context window which is used to give the model inputs of the same magnitude and then used to rescale back to the original magnitude. -
static_features: (
torch.FloatTensorof shape(batch_size, feature size), optional) — Static features of each time series’ in a batch which are copied to the covariates at inference time.
The AutoformerModel 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.
Examples:
>>> from huggingface_hub import hf_hub_download
>>> import torch
>>> from transformers import AutoformerModel
>>> file = hf_hub_download(
... repo_id="hf-internal-testing/tourism-monthly-batch", filename="train-batch.pt", repo_type="dataset"
... )
>>> batch = torch.load(file)
>>> model = AutoformerModel.from_pretrained("huggingface/autoformer-tourism-monthly")
>>> # during training, one provides both past and future values
>>> # as well as possible additional features
>>> outputs = model(
... past_values=batch["past_values"],
... past_time_features=batch["past_time_features"],
... past_observed_mask=batch["past_observed_mask"],
... static_categorical_features=batch["static_categorical_features"],
... future_values=batch["future_values"],
... future_time_features=batch["future_time_features"],
... )
>>> last_hidden_state = outputs.last_hidden_stateAutoformerForPrediction
class transformers.AutoformerForPrediction
< source >( config: AutoformerConfig )
Parameters
- config (AutoformerConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.
The Autoformer Model with a distribution head on top for time-series forecasting. This model inherits from PreTrainedModel. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.)
This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.
forward
< source >( past_values: Tensor past_time_features: Tensor past_observed_mask: Tensor static_categorical_features: Optional = None static_real_features: Optional = None future_values: Optional = None future_time_features: Optional = None future_observed_mask: Optional = None decoder_attention_mask: Optional = None head_mask: Optional = None decoder_head_mask: Optional = None cross_attn_head_mask: Optional = None encoder_outputs: Optional = None past_key_values: Optional = None output_hidden_states: Optional = None output_attentions: Optional = None use_cache: Optional = None return_dict: Optional = None ) → transformers.modeling_outputs.Seq2SeqTSPredictionOutput or tuple(torch.FloatTensor)
Parameters
- past_values (
torch.FloatTensorof shape(batch_size, sequence_length)) — Past values of the time series, that serve as context in order to predict the future. These values may contain lags, i.e. additional values from the past which are added in order to serve as “extra context”. Thepast_valuesis what the Transformer encoder gets as input (with optional additional features, such asstatic_categorical_features,static_real_features,past_time_features).The sequence length here is equal to
context_length+max(config.lags_sequence).Missing values need to be replaced with zeros.
- past_time_features (
torch.FloatTensorof shape(batch_size, sequence_length, num_features), optional) — Optional time features, which the model internally will add topast_values. These could be things like “month of year”, “day of the month”, etc. encoded as vectors (for instance as Fourier features). These could also be so-called “age” features, which basically help the model know “at which point in life” a time-series is. Age features have small values for distant past time steps and increase monotonically the more we approach the current time step.These features serve as the “positional encodings” of the inputs. So contrary to a model like BERT, where the position encodings are learned from scratch internally as parameters of the model, the Time Series Transformer requires to provide additional time features.
The Autoformer only learns additional embeddings for
static_categorical_features. - past_observed_mask (
torch.BoolTensorof shape(batch_size, sequence_length), optional) — Boolean mask to indicate whichpast_valueswere observed and which were missing. Mask values selected in[0, 1]:- 1 for values that are observed,
- 0 for values that are missing (i.e. NaNs that were replaced by zeros).
- static_categorical_features (
torch.LongTensorof shape(batch_size, number of static categorical features), optional) — Optional static categorical features for which the model will learn an embedding, which it will add to the values of the time series.Static categorical features are features which have the same value for all time steps (static over time).
A typical example of a static categorical feature is a time series ID.
- static_real_features (
torch.FloatTensorof shape(batch_size, number of static real features), optional) — Optional static real features which the model will add to the values of the time series.Static real features are features which have the same value for all time steps (static over time).
A typical example of a static real feature is promotion information.
- future_values (
torch.FloatTensorof shape(batch_size, prediction_length)) — Future values of the time series, that serve as labels for the model. Thefuture_valuesis what the Transformer needs to learn to output, given thepast_values.See the demo notebook and code snippets for details.
Missing values need to be replaced with zeros.
- future_time_features (
torch.FloatTensorof shape(batch_size, prediction_length, num_features), optional) — Optional time features, which the model internally will add tofuture_values. These could be things like “month of year”, “day of the month”, etc. encoded as vectors (for instance as Fourier features). These could also be so-called “age” features, which basically help the model know “at which point in life” a time-series is. Age features have small values for distant past time steps and increase monotonically the more we approach the current time step.These features serve as the “positional encodings” of the inputs. So contrary to a model like BERT, where the position encodings are learned from scratch internally as parameters of the model, the Time Series Transformer requires to provide additional features.
The Autoformer only learns additional embeddings for
static_categorical_features. - attention_mask (
torch.Tensorof shape(batch_size, sequence_length), optional) — Mask to avoid performing attention on certain token indices. Mask values selected in[0, 1]:- 1 for tokens that are not masked,
- 0 for tokens that are masked.
- decoder_attention_mask (
torch.LongTensorof shape(batch_size, target_sequence_length), optional) — Mask to avoid performing attention on certain token indices. By default, a causal mask will be used, to make sure the model can only look at previous inputs in order to predict the future. - head_mask (
torch.Tensorof shape(encoder_layers, encoder_attention_heads), optional) — Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in[0, 1]:- 1 indicates the head is not masked,
- 0 indicates the head is masked.
- decoder_head_mask (
torch.Tensorof shape(decoder_layers, decoder_attention_heads), optional) — Mask to nullify selected heads of the attention modules in the decoder. Mask values selected in[0, 1]:- 1 indicates the head is not masked,
- 0 indicates the head is masked.
- cross_attn_head_mask (
torch.Tensorof shape(decoder_layers, decoder_attention_heads), optional) — Mask to nullify selected heads of the cross-attention modules. Mask values selected in[0, 1]:- 1 indicates the head is not masked,
- 0 indicates the head is masked.
- encoder_outputs (
tuple(tuple(torch.FloatTensor), optional) — Tuple consists oflast_hidden_state,hidden_states(optional) andattentions(optional)last_hidden_stateof shape(batch_size, sequence_length, hidden_size)(optional) 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, returned whenuse_cache=Trueis passed or whenconfig.use_cache=True) — Tuple oftuple(torch.FloatTensor)of lengthconfig.n_layers, with each tuple having 2 tensors of shape(batch_size, num_heads, sequence_length, embed_size_per_head)) and 2 additional tensors of shape(batch_size, num_heads, encoder_sequence_length, embed_size_per_head).Contains pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used (see
past_key_valuesinput) to speed up sequential decoding.If
past_key_valuesare used, the user can optionally input only the lastdecoder_input_ids(those that don’t have their past key value states given to this model) of shape(batch_size, 1)instead of alldecoder_input_idsof shape(batch_size, sequence_length). - inputs_embeds (
torch.FloatTensorof shape(batch_size, sequence_length, hidden_size), optional) — Optionally, instead of passinginput_idsyou can choose to directly pass an embedded representation. This is useful if you want more control over how to convertinput_idsindices into associated vectors than the model’s internal embedding lookup matrix. - use_cache (
bool, optional) — If set toTrue,past_key_valueskey value states are returned and can be used to speed up decoding (seepast_key_values). - output_attentions (
bool, optional) — Whether or not to return the attentions tensors of all attention layers. Seeattentionsunder returned tensors for more detail. - output_hidden_states (
bool, optional) — Whether or not to return the hidden states of all layers. Seehidden_statesunder returned tensors for more detail. - return_dict (
bool, optional) — Whether or not to return a ModelOutput instead of a plain tuple.
Returns
transformers.modeling_outputs.Seq2SeqTSPredictionOutput or tuple(torch.FloatTensor)
A transformers.modeling_outputs.Seq2SeqTSPredictionOutput or a tuple of
torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various
elements depending on the configuration (AutoformerConfig) and inputs.
-
loss (
torch.FloatTensorof shape(1,), optional, returned when afuture_valuesis provided) — Distributional loss. -
params (
torch.FloatTensorof shape(batch_size, num_samples, num_params)) — Parameters of the chosen distribution. -
past_key_values (
tuple(tuple(torch.FloatTensor)), optional, returned whenuse_cache=Trueis passed or whenconfig.use_cache=True) — Tuple oftuple(torch.FloatTensor)of lengthconfig.n_layers, with each tuple having 2 tensors of shape(batch_size, num_heads, sequence_length, embed_size_per_head)) and 2 additional tensors of shape(batch_size, num_heads, encoder_sequence_length, embed_size_per_head).Contains pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used (see
past_key_valuesinput) to speed up sequential decoding. -
decoder_hidden_states (
tuple(torch.FloatTensor), optional, returned whenoutput_hidden_states=Trueis passed or whenconfig.output_hidden_states=True) — Tuple oftorch.FloatTensor(one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape(batch_size, sequence_length, hidden_size).Hidden-states of the decoder at the output of each layer plus the initial embedding outputs.
-
decoder_attentions (
tuple(torch.FloatTensor), optional, returned whenoutput_attentions=Trueis passed or whenconfig.output_attentions=True) — Tuple oftorch.FloatTensor(one for each layer) of shape(batch_size, num_heads, sequence_length, sequence_length).Attentions weights of the decoder, after the attention softmax, used to compute the weighted average in the self-attention heads.
-
cross_attentions (
tuple(torch.FloatTensor), optional, returned whenoutput_attentions=Trueis passed or whenconfig.output_attentions=True) — Tuple oftorch.FloatTensor(one for each layer) of shape(batch_size, num_heads, sequence_length, sequence_length).Attentions weights of the decoder’s cross-attention layer, after the attention softmax, used to compute the weighted average in the cross-attention heads.
-
encoder_last_hidden_state (
torch.FloatTensorof shape(batch_size, sequence_length, hidden_size), optional) — Sequence of hidden-states at the output of the last layer of the encoder of the model. -
encoder_hidden_states (
tuple(torch.FloatTensor), optional, returned whenoutput_hidden_states=Trueis passed or whenconfig.output_hidden_states=True) — Tuple oftorch.FloatTensor(one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape(batch_size, sequence_length, hidden_size).Hidden-states of the encoder at the output of each layer plus the initial embedding outputs.
-
encoder_attentions (
tuple(torch.FloatTensor), optional, returned whenoutput_attentions=Trueis passed or whenconfig.output_attentions=True) — Tuple oftorch.FloatTensor(one for each layer) of shape(batch_size, num_heads, sequence_length, sequence_length).Attentions weights of the encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.
-
loc (
torch.FloatTensorof shape(batch_size,)or(batch_size, input_size), optional) — Shift values of each time series’ context window which is used to give the model inputs of the same magnitude and then used to shift back to the original magnitude. -
scale (
torch.FloatTensorof shape(batch_size,)or(batch_size, input_size), optional) — Scaling values of each time series’ context window which is used to give the model inputs of the same magnitude and then used to rescale back to the original magnitude. -
static_features (
torch.FloatTensorof shape(batch_size, feature size), optional) — Static features of each time series’ in a batch which are copied to the covariates at inference time.
The AutoformerForPrediction 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.
Examples:
>>> from huggingface_hub import hf_hub_download
>>> import torch
>>> from transformers import AutoformerForPrediction
>>> file = hf_hub_download(
... repo_id="hf-internal-testing/tourism-monthly-batch", filename="train-batch.pt", repo_type="dataset"
... )
>>> batch = torch.load(file)
>>> model = AutoformerForPrediction.from_pretrained("huggingface/autoformer-tourism-monthly")
>>> # during training, one provides both past and future values
>>> # as well as possible additional features
>>> outputs = model(
... past_values=batch["past_values"],
... past_time_features=batch["past_time_features"],
... past_observed_mask=batch["past_observed_mask"],
... static_categorical_features=batch["static_categorical_features"],
... future_values=batch["future_values"],
... future_time_features=batch["future_time_features"],
... )
>>> loss = outputs.loss
>>> loss.backward()
>>> # during inference, one only provides past values
>>> # as well as possible additional features
>>> # the model autoregressively generates future values
>>> outputs = model.generate(
... past_values=batch["past_values"],
... past_time_features=batch["past_time_features"],
... past_observed_mask=batch["past_observed_mask"],
... static_categorical_features=batch["static_categorical_features"],
... future_time_features=batch["future_time_features"],
... )
>>> mean_prediction = outputs.sequences.mean(dim=1)The AutoformerForPrediction can also use static_real_features. To do so, set num_static_real_features in AutoformerConfig based on number of such features in the dataset (in case of tourism_monthly dataset it
is equal to 1), initialize the model and call as shown below:
>>> from huggingface_hub import hf_hub_download
>>> import torch
>>> from transformers import AutoformerConfig, AutoformerForPrediction
>>> file = hf_hub_download(
... repo_id="hf-internal-testing/tourism-monthly-batch", filename="train-batch.pt", repo_type="dataset"
... )
>>> batch = torch.load(file)
>>> # check number of static real features
>>> num_static_real_features = batch["static_real_features"].shape[-1]
>>> # load configuration of pretrained model and override num_static_real_features
>>> configuration = AutoformerConfig.from_pretrained(
... "huggingface/autoformer-tourism-monthly",
... num_static_real_features=num_static_real_features,
... )
>>> # we also need to update feature_size as it is not recalculated
>>> configuration.feature_size += num_static_real_features
>>> model = AutoformerForPrediction(configuration)
>>> outputs = model(
... past_values=batch["past_values"],
... past_time_features=batch["past_time_features"],
... past_observed_mask=batch["past_observed_mask"],
... static_categorical_features=batch["static_categorical_features"],
... static_real_features=batch["static_real_features"],
... future_values=batch["future_values"],
... future_time_features=batch["future_time_features"],
... )