Models used to benchmark forecasting methods on social time-series dataset for Lead-Lag dynamics.

• partial repo, Time-MoE foundation model benchmarks are in another repository.

• to run transformer, MLP and LSTM models call main_bench.py with desired set of args. You will need to do some data preprocessing on the released data before running this, limited to changing keys and extracting labels.

• the models are in baseline_models. Ignore train.py, this model is unrelated to the currrent released paper!

• will consolidate and add documentation if time permits!