The repo is the official implementation for the paper: MambaDBF: Dual-Branch Mamba with FFN for Time Series Forecasting
Key codes:
- For the architecture design of MambaDBF, please refer primarily to
models/MambaDBF.py. - For MambaFFN, please refer mainly to
layers/Mambaffn.py. - For Weighted Signal Decay Loss (EWSDL), please focus on the
exp/Exp_Long_Term_Forecast_EWSDL.py.
-
Install Python 3.8. For convenience, execute the following command.
pip install -r requirements.txt
-
For setting up the Mamba environment, please refer to https://github.com/state-spaces/mamba. Here is a simple instruction on Linux system,
pip install causal-conv1d>=1.2.0 pip install mamba-ssm -
Train and evaluate model. We provide the experiment scripts for all benchmarks under the folder
./scripts/. You can reproduce the experiment results as the following examples:sh ./scripts/MambaDBF_scripts/MambaDBF_Weather.sh