jaeho3690 / LTSF-SystemAnalysis

Instruction

1. We only need ETT-small. Place your data on data/ETT-small.
2. python=3.9 is recommended. others need to be checked.
3. Install pytorch, pytorch-lightning==1.6.5 and hydra.
4. First Install pytorch and then do the next installations.
5. Save ETT-small data on data/ directory

1 - CUDA 11.6

pip install torch==1.10.0+cu111
pip install pytorch-lightning==1.6.5

2 - CUDA 10.1

pip install torch==1.8.1
pip install torchvision==0.9.1
pip install pytorch-lightning==1.6.5
pip install hydra-core==1.2.0
pip install hydra-colorlog==1.2.0
pip install setuptools==59.5.0

RUN EXPERIMENTS

PLEASE RUN THEM one by one!! \ see if this dev file first works. You should see the Fast dev run works printed out, this may take a few minutes

sh scripts/fast_dev_run.sh 

run experiments The outputs are stored in the outputs directory

sh run_all_exp.sh

After EXPERIMENTS

run the jupyter file notebook/combine_all_files.ipynb, send me the all_df.