A unified benchmarking framework for evaluating Time Series Foundation Models (TSFMs) on carbon intensity forecasting.
cd open_tsfm_carbon
nohup bash ./scripts/run_chronos.sh > ./logs/chronos.log 2>&1 &
nohup bash ./scripts/run_moment.sh > ./logs/moment.log 2>&1 &
nohup bash ./scripts/run_timesfm.sh > ./logs/timesfm.log 2>&1 &
nohup bash ./scripts/run_uni2ts.sh > ./logs/uni2ts.log 2>&1 &
nohup bash ./scripts/run_auroarima.sh > ./logs/auroarima.log 2>&1 &
bash ./time_series_library/scripts/long_term_forecast/ETT_script/run_all_regions_TimeXer.sh
python ./EWMA/ewmaCIForecast.py
# Create environment
virtualenv chronos -p python3.10
source chronos/bin/activate
# Install packages
pip install -r requirements.txt
python3.10 -m pip install git+https://github.com/amazon-science/chronos-forecasting.git
You can find it in this repo.
⚠️ Only Python ≥ 3.10 is supported.
# Create environment
virtualenv moment -p python3.10
source moment/bin/activate
# Install packages
pip install -r requirements.txt
python3.10 -m pip install git+https://github.com/moment-timeseries-foundation-model/moment.git
# Clone repo
git clone https://github.com/SalesforceAIResearch/uni2ts.git
cd uni2ts
# Create environment
virtualenv uni2ts -p python3.10
source uni2ts/bin/activate
# Install
pip install -e '.[notebook]'
pip install -r requirements.txt
Refer to the official repo for full instructions.
✅ As of 2024-09-30, you can install via:
pip install timesfm
Repo: https://github.com/thuml/Time-Series-Library
Follow the instructions in the official repo.