iTransformer
0.8.0
Implementación de iTransformer: pronóstico de series temporales SOTA utilizando redes de atención, fuera del grupo Tsinghua / Ant
Todo lo que queda son datos tabulares (xgboost sigue siendo el campeón aquí) antes de que uno pueda realmente declarar "Atención es todo lo que necesita"
Antes de que Apple consiga que los autores cambien el nombre.
¡La implementación oficial ha sido publicada aquí!
EstabilidadAI y ? Huggingface por el generoso patrocinio, así como a mis otros patrocinadores, por brindarme la independencia para abrir las técnicas actuales de inteligencia artificial.
Greg DeVos por compartir los experimentos que realizó en iTransformer y algunas de las variantes improvisadas.
$ pip install iTransformer import torch
from iTransformer import iTransformer
# using solar energy settings
model = iTransformer (
num_variates = 137 ,
lookback_len = 96 , # or the lookback length in the paper
dim = 256 , # model dimensions
depth = 6 , # depth
heads = 8 , # attention heads
dim_head = 64 , # head dimension
pred_length = ( 12 , 24 , 36 , 48 ), # can be one prediction, or many
num_tokens_per_variate = 1 , # experimental setting that projects each variate to more than one token. the idea is that the network can learn to divide up into time tokens for more granular attention across time. thanks to flash attention, you should be able to accommodate long sequence lengths just fine
use_reversible_instance_norm = True # use reversible instance normalization, proposed here https://openreview.net/forum?id=cGDAkQo1C0p . may be redundant given the layernorms within iTransformer (and whatever else attention learns emergently on the first layer, prior to the first layernorm). if i come across some time, i'll gather up all the statistics across variates, project them, and condition the transformer a bit further. that makes more sense
)
time_series = torch . randn ( 2 , 96 , 137 ) # (batch, lookback len, variates)
preds = model ( time_series )
# preds -> Dict[int, Tensor[batch, pred_length, variate]]
# -> (12: (2, 12, 137), 24: (2, 24, 137), 36: (2, 36, 137), 48: (2, 48, 137)) Para una versión improvisada que brinda atención granular a través de tokens de tiempo (así como los tokens originales por variable), simplemente importe iTransformer2D y configure los num_time_tokens adicionales.
Actualización: ¡Funciona! ¡Gracias a Greg DeVos por realizar el experimento aquí!
Actualización 2: Recibí un correo electrónico. Sí, eres libre de escribir un artículo sobre esto, si la arquitectura se adapta a tu problema. No tengo piel en el juego
import torch
from iTransformer import iTransformer2D
# using solar energy settings
model = iTransformer2D (
num_variates = 137 ,
num_time_tokens = 16 , # number of time tokens (patch size will be (look back length // num_time_tokens))
lookback_len = 96 , # the lookback length in the paper
dim = 256 , # model dimensions
depth = 6 , # depth
heads = 8 , # attention heads
dim_head = 64 , # head dimension
pred_length = ( 12 , 24 , 36 , 48 ), # can be one prediction, or many
use_reversible_instance_norm = True # use reversible instance normalization
)
time_series = torch . randn ( 2 , 96 , 137 ) # (batch, lookback len, variates)
preds = model ( time_series )
# preds -> Dict[int, Tensor[batch, pred_length, variate]]
# -> (12: (2, 12, 137), 24: (2, 24, 137), 36: (2, 36, 137), 48: (2, 48, 137)) Un iTransformer pero también con tokens de Fourier (la FFT de la serie temporal se proyecta en tokens propios y se atiende junto con los tokens variados, separados al final)
import torch
from iTransformer import iTransformerFFT
# using solar energy settings
model = iTransformerFFT (
num_variates = 137 ,
lookback_len = 96 , # or the lookback length in the paper
dim = 256 , # model dimensions
depth = 6 , # depth
heads = 8 , # attention heads
dim_head = 64 , # head dimension
pred_length = ( 12 , 24 , 36 , 48 ), # can be one prediction, or many
num_tokens_per_variate = 1 , # experimental setting that projects each variate to more than one token. the idea is that the network can learn to divide up into time tokens for more granular attention across time. thanks to flash attention, you should be able to accommodate long sequence lengths just fine
use_reversible_instance_norm = True # use reversible instance normalization, proposed here https://openreview.net/forum?id=cGDAkQo1C0p . may be redundant given the layernorms within iTransformer (and whatever else attention learns emergently on the first layer, prior to the first layernorm). if i come across some time, i'll gather up all the statistics across variates, project them, and condition the transformer a bit further. that makes more sense
)
time_series = torch . randn ( 2 , 96 , 137 ) # (batch, lookback len, variates)
preds = model ( time_series )
# preds -> Dict[int, Tensor[batch, pred_length, variate]]
# -> (12: (2, 12, 137), 24: (2, 24, 137), 36: (2, 36, 137), 48: (2, 48, 137)) @misc { liu2023itransformer ,
title = { iTransformer: Inverted Transformers Are Effective for Time Series Forecasting } ,
author = { Yong Liu and Tengge Hu and Haoran Zhang and Haixu Wu and Shiyu Wang and Lintao Ma and Mingsheng Long } ,
year = { 2023 } ,
eprint = { 2310.06625 } ,
archivePrefix = { arXiv } ,
primaryClass = { cs.LG }
} @misc { shazeer2020glu ,
title = { GLU Variants Improve Transformer } ,
author = { Noam Shazeer } ,
year = { 2020 } ,
url = { https://arxiv.org/abs/2002.05202 }
} @misc { burtsev2020memory ,
title = { Memory Transformer } ,
author = { Mikhail S. Burtsev and Grigory V. Sapunov } ,
year = { 2020 } ,
eprint = { 2006.11527 } ,
archivePrefix = { arXiv } ,
primaryClass = { cs.CL }
} @inproceedings { Darcet2023VisionTN ,
title = { Vision Transformers Need Registers } ,
author = { Timoth'ee Darcet and Maxime Oquab and Julien Mairal and Piotr Bojanowski } ,
year = { 2023 } ,
url = { https://api.semanticscholar.org/CorpusID:263134283 }
} @inproceedings { dao2022flashattention ,
title = { Flash{A}ttention: Fast and Memory-Efficient Exact Attention with {IO}-Awareness } ,
author = { Dao, Tri and Fu, Daniel Y. and Ermon, Stefano and Rudra, Atri and R{'e}, Christopher } ,
booktitle = { Advances in Neural Information Processing Systems } ,
year = { 2022 }
} @Article { AlphaFold2021 ,
author = { Jumper, John and Evans, Richard and Pritzel, Alexander and Green, Tim and Figurnov, Michael and Ronneberger, Olaf and Tunyasuvunakool, Kathryn and Bates, Russ and {v{Z}}{'i}dek, Augustin and Potapenko, Anna and Bridgland, Alex and Meyer, Clemens and Kohl, Simon A A and Ballard, Andrew J and Cowie, Andrew and Romera-Paredes, Bernardino and Nikolov, Stanislav and Jain, Rishub and Adler, Jonas and Back, Trevor and Petersen, Stig and Reiman, David and Clancy, Ellen and Zielinski, Michal and Steinegger, Martin and Pacholska, Michalina and Berghammer, Tamas and Bodenstein, Sebastian and Silver, David and Vinyals, Oriol and Senior, Andrew W and Kavukcuoglu, Koray and Kohli, Pushmeet and Hassabis, Demis } ,
journal = { Nature } ,
title = { Highly accurate protein structure prediction with {AlphaFold} } ,
year = { 2021 } ,
doi = { 10.1038/s41586-021-03819-2 } ,
note = { (Accelerated article preview) } ,
} @inproceedings { kim2022reversible ,
title = { Reversible Instance Normalization for Accurate Time-Series Forecasting against Distribution Shift } ,
author = { Taesung Kim and Jinhee Kim and Yunwon Tae and Cheonbok Park and Jang-Ho Choi and Jaegul Choo } ,
booktitle = { International Conference on Learning Representations } ,
year = { 2022 } ,
url = { https://openreview.net/forum?id=cGDAkQo1C0p }
} @inproceedings { Katsch2023GateLoopFD ,
title = { GateLoop: Fully Data-Controlled Linear Recurrence for Sequence Modeling } ,
author = { Tobias Katsch } ,
year = { 2023 } ,
url = { https://api.semanticscholar.org/CorpusID:265018962 }
} @article { Zhou2024ValueRL ,
title = { Value Residual Learning For Alleviating Attention Concentration In Transformers } ,
author = { Zhanchao Zhou and Tianyi Wu and Zhiyun Jiang and Zhenzhong Lan } ,
journal = { ArXiv } ,
year = { 2024 } ,
volume = { abs/2410.17897 } ,
url = { https://api.semanticscholar.org/CorpusID:273532030 }
} @article { Zhu2024HyperConnections ,
title = { Hyper-Connections } ,
author = { Defa Zhu and Hongzhi Huang and Zihao Huang and Yutao Zeng and Yunyao Mao and Banggu Wu and Qiyang Min and Xun Zhou } ,
journal = { ArXiv } ,
year = { 2024 } ,
volume = { abs/2409.19606 } ,
url = { https://api.semanticscholar.org/CorpusID:272987528 }
}
