SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

time series forecasting model (#17965)

Browse Source 
* initial files

* initial model via cli

* typos

* make a start on the model config

* ready with configuation

* remove tokenizer ref.

* init the transformer

* added initial model forward to return dec_output

* require gluonts

* update dep. ver table and add as extra

* fixed typo

* add type for prediction_length

* use num_time_features

* use config

* more config

* typos

* opps another typo

* freq can be none

* default via transformation is 1

* initial transformations

* fix imports

* added transform_start_field

* add helper to create pytorch dataloader

* added inital val and test data loader

* added initial distr head and loss

* training working

* remove TimeSeriesTransformerTokenizer

Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update src/transformers/__init__.py

Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update src/transformers/models/time_series_transformer/__init__.py

Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* fixed copyright

* removed docs

* remove time series tokenizer

* fixed docs

* fix text

* fix second

* fix default

* fix order

* use config directly

* undo change

* fix comment

* fix year

* fix import

* add additional arguments for training vs. test

* initial greedy inference loop

* fix inference

* comment out token inputs to enc dec

* Use HF encoder/decoder

* fix inference

* Use Seq2SeqTSModelOutput output

* return Seq2SeqTSPredictionOutput

* added default arguments

* fix return_dict true

* scale is a tensor

* output static_features for inference

* clean up some unused bits

* fixed typo

* set return_dict if none

* call model once for both train/predict

* use cache if future_target is none

* initial generate func

* generate arguments

* future_time_feat is required

* return SampleTSPredictionOutput

* removed unneeded classes

* fix when params is none

* fix return dict

* fix num_attention_heads

* fix arguments

* remove unused shift_tokens_right

* add different dropout configs

* implement FeatureEmbedder, Scaler and weighted_average

* remove gluonts dependency

* fix class names

* avoid _variable names

* remove gluonts dependency

* fix imports

* remove gluonts from configuration

* fix docs

* fixed typo

* move utils to examples

* add example requirements

* config has no freq

* initial run_ts_no_trainer

* remove from ignore

* fix output_attentions and removed unsued getters/setters

* removed unsed tests

* add dec seq len

* add test_attention_outputs

* set has_text_modality=False

* add config attribute_map

* make style

* make fix-copies

* add encoder_outputs to TimeSeriesTransformerForPrediction forward

* Improve docs, add model to README

* added test_forward_signature

* More improvements

* Add more copied from

* Fix README

* Fix remaining quality issues

* updated encoder and decoder

* fix generate

* output_hidden_states and use_cache are optional

* past key_values returned too

* initialize weights of distribution_output module

* fixed more tests

* update test_forward_signature

* fix return_dict outputs

* Update src/transformers/models/time_series_transformer/configuration_time_series_transformer.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/time_series_transformer/configuration_time_series_transformer.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/time_series_transformer/configuration_time_series_transformer.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/time_series_transformer/configuration_time_series_transformer.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/time_series_transformer/modeling_time_series_transformer.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/time_series_transformer/modeling_time_series_transformer.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/time_series_transformer/modeling_time_series_transformer.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* removed commented out tests

* added neg. bin and normal output

* Update src/transformers/models/time_series_transformer/configuration_time_series_transformer.py

Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* move to one line

* Add docstrings

* Update src/transformers/models/time_series_transformer/configuration_time_series_transformer.py

Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* add try except for assert and raise

* try and raise exception

* fix the documentation formatting

* fix assert call

* fix docstring formatting

* removed input_ids from DOCSTRING

* Update input docstring

* Improve variable names

* Update order of inputs

* Improve configuration

* Improve variable names

* Improve docs

* Remove key_length from tests

* Add extra docs

* initial unittests

* added test_inference_no_head test

* added test_inference_head

* add test_seq_to_seq_generation

* make style

* one line

* assert mean prediction

* removed comments

* Update src/transformers/models/time_series_transformer/modeling_time_series_transformer.py

Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update src/transformers/models/time_series_transformer/modeling_time_series_transformer.py

Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* fix order of args

* make past_observed_mask optional as well

* added Amazon license header

* updated utils with new fieldnames

* make style

* cleanup

* undo position of past_observed_mask

* fix import

* typo

* more typo

* rename example files

* remove example for now

* Update docs/source/en/_toctree.yml

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/time_series_transformer/configuration_time_series_transformer.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/time_series_transformer/modeling_time_series_transformer.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/time_series_transformer/modeling_time_series_transformer.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update modeling_time_series_transformer.py

fix style

* fixed typo

* fix typo and grammer

* fix style

Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>
Co-authored-by: NielsRogge <niels.rogge1@gmail.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
This commit is contained in:
Kashif Rasul
2022-09-30 21:32:59 +02:00
committed by GitHub
parent cfb777f27c
commit 5cd16f01db
19 changed files with 2828 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

73
docs/source/en/model_doc/time_series_transformer.mdx Normal file
View File

@@ -0,0 +1,73 @@
<!--Copyright 2022 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# Time Series Transformer
<Tip>
This is a recently introduced model so the API hasn't been tested extensively. There may be some bugs or slight
breaking changes to fix it in the future. If you see something strange, file a [Github Issue](https://github.com/huggingface/transformers/issues/new?assignees=&labels=&template=bug-report.md&title).
</Tip>
## Overview
The Time Series Transformer model is a vanilla encoder-decoder Transformer for time series forecasting.
Tips:
- Similar to other models in the library, [`TimeSeriesTransformerModel`] is the raw Transformer without any head on top, and [`TimeSeriesTransformerForPrediction`]
adds a distribution head on top of the former, which can be used for time-series forecasting. Note that this is a so-called probabilistic forecasting model, not a
point forecasting model. This means that the model learns a distribution, from which one can sample. The model doesn't directly output values.
- [`TimeSeriesTransformerForPrediction`] consists of 2 blocks: an encoder, which takes a `context_length` of time series values as input (called `past_values`),
and a decoder, which predicts a `prediction_length` of time series values into the future (called `future_values`). During training, one needs to provide
pairs of (`past_values` and `future_values`) to the model.
- In addition to the raw (`past_values` and `future_values`), one typically provides additional features to the model. These can be the following:
- `past_time_features`: temporal features which the model will add to `past_values`. These serve as "positional encodings" for the Transformer encoder.
Examples are "day of the month", "month of the year", etc. as scalar values (and then stacked together as a vector).
e.g. if a given time-series value was obtained on the 11th of August, then one could have [11, 8] as time feature vector (11 being "day of the month", 8 being "month of the year").
- `future_time_features`: temporal features which the model will add to `future_values`. These serve as "positional encodings" for the Transformer decoder.
Examples are "day of the month", "month of the year", etc. as scalar values (and then stacked together as a vector).
e.g. if a given time-series value was obtained on the 11th of August, then one could have [11, 8] as time feature vector (11 being "day of the month", 8 being "month of the year").
- `static_categorical_features`: categorical features which are static over time (i.e., have the same value for all `past_values` and `future_values`).
An example here is the store ID or region ID that identifies a given time-series.
Note that these features need to be known for ALL data points (also those in the future).
- `static_real_features`: real-valued features which are static over time (i.e., have the same value for all `past_values` and `future_values`).
An example here is the image representation of the product for which you have the time-series values (like the [ResNet](resnet) embedding of a "shoe" picture,
if your time-series is about the sales of shoes).
Note that these features need to be known for ALL data points (also those in the future).
- The model is trained using "teacher-forcing", similar to how a Transformer is trained for machine translation. This means that, during training, one shifts the
`future_values` one position to the right as input to the decoder, prepended by the last value of `past_values`. At each time step, the model needs to predict the
next target. So the set-up of training is similar to a GPT model for language, except that there's no notion of `decoder_start_token_id` (we just use the last value
of the context as initial input for the decoder).
- At inference time, we give the final value of the `past_values` as input to the decoder. Next, we can sample from the model to make a prediction at the next time step,
which is then fed to the decoder in order to make the next prediction (also called autoregressive generation).
This model was contributed by [kashif](<https://huggingface.co/kashif).
## TimeSeriesTransformerConfig
[[autodoc]] TimeSeriesTransformerConfig
## TimeSeriesTransformerModel
[[autodoc]] TimeSeriesTransformerModel
- forward
## TimeSeriesTransformerForPrediction
[[autodoc]] TimeSeriesTransformerForPrediction
- forward