{"id":20349976,"url":"https://github.com/cure-lab/scinet","last_synced_at":"2025-04-04T08:08:53.194Z","repository":{"id":38110004,"uuid":"405647009","full_name":"cure-lab/SCINet","owner":"cure-lab","description":"The GitHub repository for the paper: “Time Series is a Special Sequence: Forecasting with Sample Convolution and Interaction“. (NeurIPS 2022)","archived":false,"fork":false,"pushed_at":"2023-07-12T01:55:06.000Z","size":536,"stargazers_count":650,"open_issues_count":25,"forks_count":127,"subscribers_count":10,"default_branch":"main","last_synced_at":"2025-03-28T07:08:25.281Z","etag":null,"topics":["deep-neural-networks","forecasting","temporal-networks","time-series","time-series-analysis","time-series-forecasting"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/cure-lab.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null}},"created_at":"2021-09-12T13:18:03.000Z","updated_at":"2025-03-26T09:37:55.000Z","dependencies_parsed_at":"2023-09-22T07:01:59.889Z","dependency_job_id":null,"html_url":"https://github.com/cure-lab/SCINet","commit_stats":{"total_commits":211,"total_committers":9,"mean_commits":"23.444444444444443","dds":0.6161137440758293,"last_synced_commit":"02e6b0af2d58243de09aaa1eac3840237b659847"},"previous_names":[],"tags_count":1,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/cure-lab%2FSCINet","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/cure-lab%2FSCINet/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/cure-lab%2FSCINet/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/cure-lab%2FSCINet/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/cure-lab","download_url":"https://codeload.github.com/cure-lab/SCINet/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":247142070,"owners_count":20890652,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["deep-neural-networks","forecasting","temporal-networks","time-series","time-series-analysis","time-series-forecasting"],"created_at":"2024-11-14T22:28:26.602Z","updated_at":"2025-04-04T08:08:53.167Z","avatar_url":"https://github.com/cure-lab.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# SCINet\n[![Arxiv link](https://img.shields.io/badge/arXiv-Time%20Series%20is%20a%20Special%20Sequence%3A%20Forecasting%20with%20Sample%20Convolution%20and%20Interaction-%23B31B1B)](https://arxiv.org/pdf/2106.09305.pdf)\n\n![state-of-the-art](https://img.shields.io/badge/-STATE--OF--THE--ART-blue?logo=Accenture\u0026labelColor=lightgrey)![pytorch](https://img.shields.io/badge/-PyTorch-%23EE4C2C?logo=PyTorch\u0026labelColor=lightgrey)\n\n\n\nThis is the original pytorch implementation for the following paper: [SCINet: Time Series Modeling and Forecasting with Sample Convolution and Interaction](https://arxiv.org/pdf/2106.09305.pdf). Alse see the [Open Review verision](https://openreview.net/pdf?id=AyajSjTAzmg).  \n\nIf you find this repository useful for your research work, please consider citing it as follows:\n\n```\n\n@article{liu2022SCINet,\n\ntitle={SCINet: Time Series Modeling and Forecasting with Sample Convolution and Interaction},\n\nauthor={Liu, Minhao and Zeng, Ailing and Chen, Muxi and Xu, Zhijian and Lai, Qiuxia and Ma, Lingna and Xu, Qiang},\n\njournal={Thirty-sixth Conference on Neural Information Processing Systems (NeurIPS), 2022},\n\nyear={2022}\n\n}\n\n```\n\n## Updates\n- [2022-09-15] SCINet has been accepted to NeurIPS 2022!\n\n- [2021-11-10] Added Reversible Instance Normalization RevIN[1] support!\n\n- [2021-09-17] SCINet v1.0 is released\n\n\n## Features\n\n- [x] Support **11** popular time-series forecasting datasets, namely Electricity Transformer Temperature (ETTh1, ETTh2 and ETTm1) , Traffic, Solar-Energy, Electricity and Exchange Rate and PeMS (PEMS03, PEMS04, PEMS07 and PEMS08), ranging from power, energy, finance and traffic domains.\n\n[comment]: \u003c\u003e (![traffic]\u0026#40;https://img.shields.io/badge/🚅-Traffic-yellow\u0026#41;)\n\n[comment]: \u003c\u003e (![electric]\u0026#40;https://img.shields.io/badge/%F0%9F%92%A1-Electricity-yellow\u0026#41;)\n\n[comment]: \u003c\u003e (![Solar Energy]\u0026#40;https://img.shields.io/badge/%F0%9F%94%86-Solar%20Energy-yellow\u0026#41;)\n\n[comment]: \u003c\u003e (![finance]\u0026#40;https://img.shields.io/badge/💵-Finance-yellow\u0026#41;)\n\n- [x] Provide all training logs.\n\n- [x] Support RevIN to handle datasets with a large train-test sample distribution gap. To activate, simply add ```--RIN True``` to the command line. [**Read more**]\u0026#40;./docs/RevIN.md\u0026#41;\n\n\n## To-do items\n\n-  Integrate GNN-based spatial models into SCINet for better performance and higher efficiency on spatial-temporal time series. Our preliminary results show that this feature could result in considerable gains on the prediction accuracy of some datasets \u0026#40;e.g., PEMSxx\u0026#41;.\n\n-  Generate probalistic forecasting results.\n\nStay tuned!\n\n\n\n\n## Used Datasets\n\n\nWe conduct the experiments on **11** popular time-series datasets, namely **Electricity Transformer Temperature (ETTh1, ETTh2 and ETTm1) ,  PeMS (PEMS03, PEMS04, PEMS07 and PEMS08) and Traffic, Solar-Energy, Electricity and Exchange Rate**, ranging from **power, energy, finance and traffic domains**. \n\n\n### Overall information of the 11 datasets\n\n| Datasets      | Variants | Timesteps | Granularity | Start time | Task Type   |\n| ------------- | -------- | --------- | ----------- | ---------- | ----------- |\n| ETTh1         | 7        | 17,420    | 1hour       | 7/1/2016   | Multi-step  |\n| ETTh2         | 7        | 17,420    | 1hour       | 7/1/2016   | Multi-step  |\n| ETTm1         | 7        | 69,680    | 15min       | 7/1/2016   | Multi-step  |\n| PEMS03        | 358      | 26,209    | 5min        | 5/1/2012   | Multi-step  |\n| PEMS04        | 307      | 16,992    | 5min        | 7/1/2017   | Multi-step  |\n| PEMS07        | 883      | 28,224    | 5min        | 5/1/2017   | Multi-step  |\n| PEMS08        | 170      | 17,856    | 5min        | 3/1/2012   | Multi-step  |\n| Traffic       | 862      | 17,544    | 1hour       | 1/1/2015   | Single-step |\n| Solar-Energy  | 137      | 52,560    | 1hour       | 1/1/2006   | Single-step |\n| Electricity   | 321      | 26,304    | 1hour       | 1/1/2012   | Single-step |\n| Exchange-Rate | 8        | 7,588     | 1hour       | 1/1/1990   | Single-step |\n\n\n## Get started\n\n### Requirements\n\nInstall the required package first:\n\n```\ncd SCINet\nconda create -n scinet python=3.8\nconda activate scinet\npip install -r requirements.txt\n```\n\n### Dataset preparation\n\nAll datasets can be downloaded [here](https://drive.google.com/drive/folders/1Gv1MXjLo5bLGep4bsqDyaNMI2oQC9GH2?usp=sharing). To prepare all dataset at one time, you can just run:\n```\nsource prepare_data.sh\n```\n [![ett](https://img.shields.io/badge/Download-ETT_Dataset-%234285F4?logo=GoogleDrive\u0026labelColor=lightgrey)](https://drive.google.com/drive/folders/1NU85EuopJNkptFroPtQVXMZE70zaBznZ)\n[![pems](https://img.shields.io/badge/Download-PeMS_Dataset-%234285F4?logo=GoogleDrive\u0026labelColor=lightgrey)](https://drive.google.com/drive/folders/17fwxGyQ3Qb0TLOalI-Y9wfgTPuXSYgiI)\n[![financial](https://img.shields.io/badge/Download-financial_Dataset-%234285F4?logo=GoogleDrive\u0026labelColor=lightgrey)](https://drive.google.com/drive/folders/12ffxwxVAGM_MQiYpIk9aBLQrb2xQupT-) \n\nThe data directory structure is shown as follows. \n```\n./\n└── datasets/\n    ├── ETT-data\n    │   ├── ETTh1.csv\n    │   ├── ETTh2.csv\n    │   └── ETTm1.csv\n    ├── financial\n    │   ├── electricity.txt\n    │   ├── exchange_rate.txt\n    │   ├── solar_AL.txt\n    │   └── traffic.txt\n    └── PEMS\n        ├── PEMS03.npz\n        ├── PEMS04.npz\n        ├── PEMS07.npz\n        └── PEMS08.npz\n```\n\n### Run training code\n\nWe follow the same settings of [StemGNN](https://github.com/microsoft/StemGNN) for PEMS 03, 04, 07, 08 datasets, [MTGNN](https://github.com/nnzhan/MTGNN) for Solar, electricity, traffic, financial datasets, [Informer](https://github.com/zhouhaoyi/Informer2020) for ETTH1, ETTH2, ETTM1 datasets. The detailed training commands are given as follows.\n\n#### For PEMS dataset (All datasets follow Input 12, Output 12):\n\npems03\n```\npython run_pems.py --dataset PEMS03 --hidden-size 0.0625 --dropout 0.25 --model_name pems03_h0.0625_dp0.25 --num_decoder_layer 2\n```\n\npems04\n```\npython run_pems.py --dataset PEMS04 --hidden-size 0.0625 --dropout 0 --model_name pems04_h0.0625_dp0\n```\n\npems07\n```\npython run_pems.py --dataset PEMS07 --hidden-size 0.03125 --dropout 0.25 --model_name pems07_h0.03125_dp0.25\n```\n\npems08\n```\npython run_pems.py --dataset PEMS08 --hidden-size 1 --dropout 0.5 --model_name pems08_h1_dp0.5\n```\n\n##### PEMS Parameter highlights\n\n| Parameter Name | Description             | Parameter in paper | Default |\n| -------------- | ----------------------- | ------------------ | ------- |\n| dataset        | Name of dataset         | N/A                | PEMS08  |\n| horizon        | Horizon                 | Horizon            | 12      |\n| window_size    | Look-back window        | Look-back window   | 12      |\n| hidden-size    | hidden expansion        | h                  | 1       |\n| levels         | SCINet block levels     | L                  | 2       |\n| stacks         | The number of SCINet block| K                | 1       |\n\n\n#### For Solar dataset:\n\npredict 3 \n```\npython run_financial.py --dataset_name solar_AL --window_size 160 --horizon 3 --hidden-size 1  --lastWeight 0.5 --stacks 2 --levels 4 --lradj 2 --lr 1e-4 --dropout 0.25 --batch_size 256 --model_name so_I160_o3_lr1e-4_bs256_dp0.25_h1_s2l4_w0.5\n```\npredict 6\n```\npython run_financial.py --dataset_name solar_AL --window_size 160 --horizon 6 --hidden-size 0.5 --lastWeight 0.5 --stacks 2 --levels 4 --lradj 2 --lr 1e-4 --dropout 0.25 --batch_size 256 --model_name so_I160_o6_lr1e-4_bs256_dp0.25_h0.5_s2l4_w0.5 \n```\npredict 12\n```\npython run_financial.py --dataset_name solar_AL --window_size 160 --horizon 12 --hidden-size 2 --lastWeight 0.5 --stacks 2 --levels 4 --lradj 2 --lr 1e-4 --dropout 0.25 --batch_size 1024 --model_name so_I160_o12_lr1e-4_bs1024_dp0.25_h2_s2l4_w0.5\n```\npredict 24\n```\npython run_financial.py --dataset_name solar_AL --window_size 160 --horizon 24 --hidden-size 1 --lastWeight 0.5 --stacks 1 --levels 4 --lradj 2 --lr 1e-4 --dropout 0.25 --batch_size 256 --model_name so_I160_o24_lr1e-4_bs256_dp0.25_h1_s1l4_w0.5\n```\n\n#### For Electricity dataset:\n\npredict 3\n``` \npython run_financial.py --dataset_name electricity --window_size 168 --horizon 3 --hidden-size 8 --single_step 1 --stacks 2 --levels 3 --lr 9e-3 --dropout 0 --batch_size 32 --model_name ele_I168_o3_lr9e-3_bs32_dp0_h8_s2l3_w0.5 --groups 321 --num_decoder_layer 2\n```\npredict 6\n```\npython run_financial.py --dataset_name electricity --window_size 168 --horizon 6 --hidden-size 8 --single_step 1 --stacks 2 --levels 3 --lr 9e-3 --dropout 0 --batch_size 32 --model_name ele_I168_o6_lr9e-3_bs32_dp0_h8_s2l3_w0.5 --groups 321 --num_decoder_layer 3\n```\npredict 12\n```\npython run_financial.py --dataset_name electricity --window_size 168 --horizon 12 --hidden-size 8 --single_step 1 --stacks 2 --levels 3 --lr 9e-3 --dropout 0 --batch_size 32 --model_name ele_I168_o12_lr9e-3_bs32_dp0_h8_s2l3_w0.5 --groups 321 --num_decoder_layer 3\n```\npredict 24\n```\npython run_financial.py --dataset_name electricity --window_size 168 --horizon 24 --hidden-size 8 --single_step 1 --stacks 2 --levels 3 --lr 9e-3 --dropout 0 --batch_size 32 --model_name ele_I168_o24_lr9e-3_bs32_dp0_h8_s2l3_w0.5 --groups 321 --num_decoder_layer 3\n```\npredict 96\n```\npython -u run_financial.py --dataset_name electricity --window_size 96 --horizon 96 --hidden-size 8 --stacks 2 --levels 3 --lr 9e-4 --dropout 0 --batch_size 32 --model_name ele_I96_o96_lr9e-4_bs32_dp0_h8_s2l3_w0.5_n4 --groups 321  --concat_len 0   --normalize 4 --long_term_forecast\n```\npredict 192\n```\npython -u run_financial.py --dataset_name electricity --window_size 96 --horizon 192 --hidden-size 8 --stacks 2 --levels 3 --lr 9e-4 --dropout 0 --batch_size 32 --model_name ele_I96_o192_lr9e-4_bs32_dp0_h8_s2l3_w0.5_n4 --groups 321  --concat_len 0   --normalize 4 --long_term_forecast\n```\npredict 336\n```\npython -u run_financial.py --dataset_name electricity --window_size 96 --horizon 336 --hidden-size 8 --stacks 2 --levels 3 --lr 9e-4 --dropout 0 --batch_size 32 --model_name ele_I168_o336_lr9e-4_bs32_dp0_h8_s2l3_w0.5_n4 --groups 321  --concat_len 0   --normalize 4 --long_term_forecast\n```\npredict 720\n```\npython -u run_financial.py --dataset_name electricity --window_size 96 --horizon 720 --hidden-size 8 --stacks 2 --levels 3 --lr 9e-4 --dropout 0 --batch_size 32 --model_name ele_I168_o24_lr9e-4_bs32_dp0_h8_s2l3_w0.5_n4 --groups 321  --concat_len 0   --normalize 4 --long_term_forecast\n```\n\n#### For Traffic dataset (warning: 20,000MiB+ memory usage!):\n\npredict 3 \n```\npython run_financial.py --dataset_name traffic --window_size 168 --horizon 3 --hidden-size 1 --single_step 1 --stacks 2 --levels 3 --lr 5e-4 --dropout 0.5 --batch_size 16 --model_name traf_I168_o3_lr5e-4_bs16_dp0.5_h1_s2l3_w1.0\n```\npredict 6\n```\npython run_financial.py --dataset_name traffic --window_size 168 --horizon 6 --hidden-size 2 --single_step 1 --stacks 1 --levels 3 --lr 5e-4 --dropout 0.25 --batch_size 16 --model_name traf_I168_o6_lr5e-4_bs16_dp0.25_h2_s1l3_w1.0\n```\npredict 12\n```\npython run_financial.py --dataset_name traffic --window_size 168 --horizon 12 --hidden-size 0.5 --single_step 1 --stacks 2 --levels 3 --lr 5e-4 --dropout 0.25 --batch_size 16 --model_name traf_I168_o12_lr5e-4_bs16_dp0.25_h0.5_s2l3_w1.0\n```\npredict 24\n```\npython run_financial.py --dataset_name traffic --window_size 168 --horizon 24 --hidden-size 2 --single_step 1 --stacks 2 --levels 2 --lr 5e-4 --dropout 0.5 --batch_size 16 --model_name traf_I168_o24_lr5e-4_bs16_dp0.5_h2_s2l2_w1.0\n```\npredict 96\n```\npython -u run_financial.py --dataset_name traffic --window_size 96 --horizon 96 --hidden-size 2 --stacks 1 --levels 3 --lr 5e-4 --dropout 0.25 --batch_size 16 --model_name traf_I96_o96_lr5e-4_bs16_dp0.25_h2_s1l3_w1.0 --normalize 4 --long_term_forecast\n```\npredict 192\n```\npython -u run_financial.py --dataset_name traffic --window_size 96 --horizon 192 --hidden-size 1 --stacks 1 --levels 3 --lr 5e-4 --dropout 0.25 --batch_size 16 --model_name traf_I96_o192_lr5e-4_bs16_dp0.25_h2_s1l3_w1.0 --normalize 4  --long_term_forecast\n```\npredict 336\n```\npython -u run_financial.py --dataset_name traffic --window_size 96 --horizon 336 --hidden-size 1 --stacks 1 --levels 3 --lr 5e-4 --dropout 0.25 --batch_size 16 --model_name traf_I96_o336_lr5e-4_bs16_dp0.25_h2_s1l3_w1.0 --normalize 4 --long_term_forecast\n```\npredict 720\n```\npython -u run_financial.py --dataset_name traffic --window_size 96 --horizon 720 --hidden-size 1 --stacks 1 --levels 3 --lr 5e-4 --dropout 0.25 --batch_size 16 --model_name traf_I96_o720_lr5e-4_bs16_dp0.25_h2_s1l3_w1.0 --normalize 4 --long_term_forecast\n```\n#### For Exchange rate dataset:\n\npredict 3 \n```\npython run_financial.py --dataset_name exchange_rate --window_size 168 --horizon 3 --hidden-size 0.125 --lastWeight 0.5 --stacks 1 --levels 3 --lr 5e-3 --dropout 0.5 --batch_size 4 --model_name ex_I168_o3_lr5e-3_bs4_dp0.5_h0.125_s1l3_w0.5 --num_decoder_layer 2 --epochs 150\n```\npredict 6\n```\npython run_financial.py --dataset_name exchange_rate --window_size 168 --horizon 6 --hidden-size 0.125 --lastWeight 0.5 --stacks 1 --levels 3 --lr 5e-3 --dropout 0.5 --batch_size 4 --model_name ex_I168_o6_lr5e-3_bs4_dp0.5_h0.125_s1l3_w0.5 --num_decoder_layer 2 --epochs 150\n```\npredict 12\n```\npython run_financial.py --dataset_name exchange_rate --window_size 168 --horizon 12 --hidden-size 0.125 --lastWeight 0.5 --stacks 1 --levels 3 --lr 5e-3 --dropout 0.5 --batch_size 4 --model_name ex_I168_o12_lr5e-3_bs4_dp0.5_h0.125_s1l3_w0.5 --num_decoder_layer 2 --epochs 150\n```\npredict 24\n```\npython run_financial.py --dataset_name exchange_rate --window_size 168 --horizon 24 --hidden-size 0.125 --lastWeight 0.5 --stacks 1 --levels 3 --lr 7e-3 --dropout 0.5 --batch_size 4 --model_name ex_I168_o24_lr7e-3_bs4_dp0.5_h0.125_s1l3_w0.5 --num_decoder_layer 2 --epochs 150\n```\npredict 96\n```\npython run_financial.py --dataset_name exchange_rate --epochs 20 --window_size 96 --horizon 96 --hidden-size 0.125 --normalize 3 --lastWeight 0.5 --stacks 1 --levels 3 --lr 5e-5  --dropout 0 --model_name final --num_decoder_layer 2 --long_term_forecast\n```\npredict 192\n```\npython run_financial.py --dataset_name exchange_rate --epochs 20 --window_size 96 --horizon 192 --hidden-size 0.125 --normalize 3 --lastWeight 0.5 --stacks 1 --levels 3 --lr 5e-5  --dropout 0 --model_name final --num_decoder_layer 2 --long_term_forecast\n```\npredict 336\n```\npython run_financial.py --dataset_name exchange_rate --epochs 20 --window_size 96 --horizon 336 --hidden-size 0.125 --normalize 3 --lastWeight 0.5 --stacks 1 --levels 3 --lr 5e-5  --dropout 0 --model_name final --num_decoder_layer 2 --long_term_forecast\n```\npredict 720\n```\npython run_financial.py --dataset_name exchange_rate --epochs 20 --window_size 96 --horizon 720 --hidden-size 0.125 --normalize 3 --lastWeight 0.5 --stacks 1 --levels 3 --lr 5e-5  --dropout 0 --model_name final --num_decoder_layer 2 --long_term_forecast\n```\n\n##### Financial Parameter highlights\n\n| Parameter Name | Description               | Parameter in paper      | Default                                |\n| -------------- | ------------------------- | ----------------------- | -------------------------------------- |\n| dataset_name   | Data name                 | N/A                     | exchange_rate                          |\n| horizon        | Horizon                   | Horizon                 | 3                                      |\n| window_size    | Look-back window          | Look-back window        | 168                                    |\n| batch_size     | Batch size                | batch size              | 8                                      |\n| lr             | Learning rate             | learning rate           | 5e-3                                   |\n| hidden-size    | hidden expansion          | h                       | 1                                      |\n| levels         | SCINet block levels       | L                       | 3                                      |\n| stacks         | The number of SCINet block| K                       | 1                                      |\n| lastweight     | Loss weight of the last frame| Loss weight ($\\lambda$) | 1.0                                 |\n\n\n#### For ETTH1 dataset:\n\nmultivariate, out 24\n```\npython run_ETTh.py --data ETTh1 --features M  --seq_len 48 --label_len 24 --pred_len 24 --hidden-size 4 --stacks 1 --levels 3 --lr 3e-3 --batch_size 8 --dropout 0.5 --model_name etth1_M_I48_O24_lr3e-3_bs8_dp0.5_h4_s1l3\n```\nmultivariate, out 48\n```\npython run_ETTh.py --data ETTh1 --features M  --seq_len 96 --label_len 48 --pred_len 48 --hidden-size 4 --stacks 1 --levels 3 --lr 0.009 --batch_size 16 --dropout 0.25 --model_name etth1_M_I96_O48_lr0.009_bs16_dp0.25_h4_s1l3\n```\nmultivariate, out 168\n```\npython run_ETTh.py --data ETTh1 --features M  --seq_len 336 --label_len 168 --pred_len 168 --hidden-size 4 --stacks 1 --levels 3 --lr 5e-4 --batch_size 32 --dropout 0.5 --model_name etth1_M_I336_O168_lr5e-4_bs32_dp0.5_h4_s1l3\n```\nmultivariate, out 336\n```\npython run_ETTh.py --data ETTh1 --features M  --seq_len 336 --label_len 336 --pred_len 336 --hidden-size 1 --stacks 1 --levels 4 --lr 1e-4 --batch_size 512 --dropout 0.5 --model_name etth1_M_I336_O336_lr1e-4_bs512_dp0.5_h1_s1l4\n```\nmultivariate, out 720\n```\npython run_ETTh.py --data ETTh1 --features M  --seq_len 736 --label_len 720 --pred_len 720 --hidden-size 1 --stacks 1 --levels 5 --lr 5e-5 --batch_size 256 --dropout 0.5 --model_name etth1_M_I736_O720_lr5e-5_bs256_dp0.5_h1_s1l5\n```\nUnivariate, out 24\n```\npython run_ETTh.py --data ETTh1 --features S  --seq_len 64 --label_len 24 --pred_len 24 --hidden-size 8 --stacks 1 --levels 3 --lr 0.007 --batch_size 64 --dropout 0.25 --model_name etth1_S_I64_O24_lr0.007_bs64_dp0.25_h8_s1l3\n```\nUnivariate, out 48\n```\npython run_ETTh.py --data ETTh1 --features S  --seq_len 720 --label_len 48 --pred_len 48 --hidden-size 4 --stacks 1 --levels 4 --lr 0.0001 --batch_size 8 --dropout 0.5 --model_name etth1_S_I720_O48_lr0.0001_bs8_dp0.5_h4_s1l4\n```\nUnivariate, out 168\n```\npython run_ETTh.py --data ETTh1 --features S  --seq_len 720 --label_len 168 --pred_len 168 --hidden-size 4 --stacks 1 --levels 4 --lr 5e-5 --batch_size 8 --dropout 0.5 --model_name etth1_S_I720_O168_lr5e-5_bs8_dp0.5_h4_s1l4\n```\nUnivariate, out 336\n```\npython run_ETTh.py --data ETTh1 --features S  --seq_len 720 --label_len 336 --pred_len 336 --hidden-size 1 --stacks 1 --levels 4 --lr 1e-3 --batch_size 128 --dropout 0.5 --model_name etth1_S_I720_O336_lr1e-3_bs128_dp0.5_h1_s1l4\n```\nUnivariate, out 720\n```\npython run_ETTh.py --data ETTh1 --features S  --seq_len 736 --label_len 720 --pred_len 720 --hidden-size 4 --stacks 1 --levels 5 --lr 1e-4 --batch_size 32 --dropout 0.5 --model_name etth1_S_I736_O720_lr1e-5_bs32_dp0.5_h4_s1l5\n```\n\n#### For ETTH2 dataset:\n\nmultivariate, out 24\n```\npython run_ETTh.py --data ETTh2 --features M  --seq_len 48 --label_len 24 --pred_len 24 --hidden-size 8 --stacks 1 --levels 3 --lr 0.007 --batch_size 16 --dropout 0.25 --model_name etth2_M_I48_O24_lr7e-3_bs16_dp0.25_h8_s1l3\n```\nmultivariate, out 48\n```\npython run_ETTh.py --data ETTh2 --features M  --seq_len 96 --label_len 48 --pred_len 48 --hidden-size 4 --stacks 1 --levels 4 --lr 0.007 --batch_size 4 --dropout 0.5 --model_name etth2_M_I96_O48_lr7e-3_bs4_dp0.5_h4_s1l4\n```\nmultivariate, out 168\n```\npython run_ETTh.py --data ETTh2 --features M  --seq_len 336 --label_len 168 --pred_len 168 --hidden-size 0.5 --stacks 1 --levels 4 --lr 5e-5 --batch_size 16 --dropout 0.5 --model_name etth2_M_I336_O168_lr5e-5_bs16_dp0.5_h0.5_s1l4\n```\nmultivariate, out 336\n```\npython run_ETTh.py --data ETTh2 --features M  --seq_len 336 --label_len 336 --pred_len 336 --hidden-size 1 --stacks 1 --levels 4 --lr 5e-5 --batch_size 128 --dropout 0.5 --model_name etth2_M_I336_O336_lr5e-5_bs128_dp0.5_h1_s1l4\n```\nmultivariate, out 720\n```\npython run_ETTh.py --data ETTh2 --features M  --seq_len 736 --label_len 720 --pred_len 720 --hidden-size 4 --stacks 1 --levels 5 --lr 1e-5 --batch_size 128 --dropout 0.5 --model_name etth2_M_I736_O720_lr1e-5_bs128_dp0.5_h4_s1l5\n```\nUnivariate, out 24\n```\npython run_ETTh.py --data ETTh2 --features S  --seq_len 48 --label_len 24 --pred_len 24 --hidden-size 4 --stacks 1 --levels 3 --lr 0.001 --batch_size 16 --dropout 0 --model_name etth2_S_I48_O24_lr1e-3_bs16_dp0_h4_s1l3\n```\nUnivariate, out 48\n```\npython run_ETTh.py --data ETTh2 --features S  --seq_len 96 --label_len 48 --pred_len 48 --hidden-size 4 --stacks 2 --levels 4 --lr 0.001 --batch_size 32 --dropout 0.5 --model_name etth2_S_I96_O48_lr1e-3_bs32_dp0.5_h4_s2l4\n```\nUnivariate, out 168\n```\npython run_ETTh.py --data ETTh2 --features S  --seq_len 336 --label_len 168 --pred_len 168 --hidden-size 4 --stacks 1 --levels 3 --lr 1e-4 --batch_size 8 --dropout 0 --model_name etth2_S_I336_O168_lr1e-4_bs8_dp0_h4_s1l3\n```\nUnivariate, out 336\n```\npython run_ETTh.py --data ETTh2 --features S  --seq_len 336 --label_len 336 --pred_len 336 --hidden-size 8 --stacks 1 --levels 3 --lr 5e-4 --batch_size 512 --dropout 0.5 --model_name etth2_S_I336_O336_lr5e-4_bs512_dp0.5_h8_s1l3\n```\nUnivariate, out 720\n```\npython run_ETTh.py --data ETTh2 --features S  --seq_len 720 --label_len 720 --pred_len 720 --hidden-size 8 --stacks 1 --levels 3 --lr 1e-5 --batch_size 128 --dropout 0.6 --model_name etth2_S_I736_O720_lr1e-5_bs128_dp0.6_h8_s1l3\n```\n\n#### For ETTM1 dataset:\n\nmultivariate, out 24\n```\npython run_ETTh.py --data ETTm1 --features M  --seq_len 48 --label_len 24 --pred_len 24 --hidden-size 4 --stacks 1 --levels 3 --lr 0.005 --batch_size 32 --dropout 0.5 --model_name ettm1_M_I48_O24_lr7e-3_bs16_dp0.25_h8_s1l3\n```\nmultivariate, out 48\n```\npython run_ETTh.py --data ETTm1 --features M  --seq_len 96 --label_len 48 --pred_len 48 --hidden-size 4 --stacks 2 --levels 4 --lr 0.001 --batch_size 16 --dropout 0.5 --model_name ettm1_M_I96_O48_lr1e-3_bs16_dp0.5_h4_s2l4\n```\nmultivariate, out 96\n```\npython run_ETTh.py --data ETTm1 --features M  --seq_len 384 --label_len 96 --pred_len 96 --hidden-size 0.5 --stacks 2 --levels 4 --lr 5e-5 --batch_size 32 --dropout 0.5 --model_name ettm1_M_I384_O96_lr5e-5_bs32_dp0.5_h0.5_s2l4\n```\nmultivariate, out 288\n```\npython run_ETTh.py --data ETTm1 --features M  --seq_len 672 --label_len 288 --pred_len 288 --hidden-size 4 --stacks 1 --levels 5 --lr 1e-5 --batch_size 32 --dropout 0.5 --model_name ettm1_M_I672_O288_lr1e-5_bs32_dp0.5_h0.5_s1l5\n```\nmultivariate, out 672\n```\npython run_ETTh.py --data ETTm1 --features M  --seq_len 672 --label_len 672 --pred_len 672 --hidden-size 4 --stacks 2 --levels 5 --lr 1e-5 --batch_size 32 --dropout 0.5 --model_name ettm1_M_I672_O672_lr1e-5_bs32_dp0.5_h4_s2l5\n```\nUnivariate, out 24\n```\npython run_ETTh.py --data ETTm1 --features S  --seq_len 96 --label_len 24 --pred_len 24 --hidden-size 4 --stacks 1 --levels 4 --lr 0.001 --batch_size 8 --dropout 0 --model_name ettm1_S_I96_O24_lr1e-3_bs8_dp0_h4_s1l4\n```\nUnivariate, out 48\n```\npython run_ETTh.py --data ETTm1 --features S  --seq_len 96 --label_len 48 --pred_len 48 --hidden-size 4 --stacks 1 --levels 3 --lr 0.0005 --batch_size 16 --dropout 0 --model_name ettm1_S_I96_O48_lr5e-4_bs16_dp0_h4_s1l3\n```\nUnivariate, out 96\n```\npython run_ETTh.py --data ETTm1 --features S  --seq_len 384 --label_len 96 --pred_len 96 --hidden-size 2 --stacks 1 --levels 4 --lr 1e-5 --batch_size 8 --dropout 0 --model_name ettm1_S_I384_O96_lr1e-5_bs8_dp0_h2_s1l4\n```\nUnivariate, out 288\n```\npython run_ETTh.py --data ETTm1 --features S  --seq_len 384 --label_len 288 --pred_len 288 --hidden-size 4 --stacks 1 --levels 4 --lr 1e-5 --batch_size 64 --dropout 0 --model_name ettm1_S_I384_O288_lr1e-5_bs64_dp0_h4_s1l4\n```\nUnivariate, out 672\n```\npython run_ETTh.py --data ETTm1 --features S  --seq_len 672 --label_len 672 --pred_len 672 --hidden-size 1 --stacks 1 --levels 5 --lr 1e-4 --batch_size 32 --model_name ettm1_S_I672_O672_lr1e-4_bs32_dp0.5_h1_s1l5\n```\n\n\n##### ETT Parameter highlights\n\n| Parameter Name | Description                  | Parameter in paper | Default                    |\n| -------------- | ---------------------------- | ------------------ | -------------------------- |\n| root_path      | The root path of subdatasets | N/A                | './datasets/ETT-data/ETT/' |\n| data           | Subdataset                   | N/A                | ETTh1                      |\n| pred_len       | Horizon                      | Horizon            | 48                         |\n| seq_len        | Look-back window             | Look-back window   | 96                         |\n| batch_size     | Batch size                   | batch size         | 32                         |\n| lr             | Learning rate                | learning rate      | 0.0001                     |\n| hidden-size    | hidden expansion             | h                  | 1                          |\n| levels         | SCINet block levels          | L                  | 3                          |\n| stacks         | The number of SCINet blocks  | K                  | 1                          |\n\n## Special Constraint\n\n- Because of the stacked binary down-sampling method that SCINet adopts, the number of levels \u0026#40;L\u0026#41; and look-back window \u0026#40;W\u0026#41; size should satisfy:)\n\n\u003cimg src=\"https://render.githubusercontent.com/render/math?math=W\\bmod{2^{L}}=0\"\u003e\n\n\u0026#40;The formula might not be shown in the darkmode Github\u0026#41;\n\n## References\n\n[1] [Reversible Instance Normalization for Accurate Time-Series Forecasting against Distribution Shift]\u0026#40;https://openreview.net/forum?id=cGDAkQo1C0p\u0026#41;\n\n## Contact\n\nIf you have any questions, feel free to contact us or post github issues. Pull requests are highly welcomed!\n\n```\n\nMinhao Liu: mhliu@cse.cuhk.edu.hk\n\nAiling Zeng: alzeng@cse.cuhk.edu.hk\n\nZhijian Xu: zjxu21@cse.cuhk.edu.hk\n\n```\n\n## Acknowledgements\n\nThank you all for your attention to our work!\n\nThis code uses ([Informer](https://github.com/zhouhaoyi/Informer2020), [MTGNN](https://github.com/nnzhan/MTGNN), [StemGNN](https://github.com/microsoft/StemGNN)) as baseline methods for comparison.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fcure-lab%2Fscinet","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fcure-lab%2Fscinet","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fcure-lab%2Fscinet/lists"}