scripts/short_term_forecast/polyswm.sh

export CUDA_VISIBLE_DEVICES=6


# model_name=
for model_name in Autoformer DLinear
do
  python -u run.py \
    --task_name short_term_forecast \
    --is_training 1 \
    --root_path /data/haofeiy2/social-world-model/data/splitted_polymarket_v2_0102 \
    --model_id poly_Daily \
    --model $model_name \
    --data poly \
    --features S \
    --e_layers 2 \
    --d_layers 1 \
    --factor 3 \
    --enc_in 1 \
    --dec_in 1 \
    --c_out 1 \
    --batch_size 128 \
    --d_model 128 \
    --des 'Exp' \
    --itr 1 \
    --seq_len 16 \
    --label_len 1 \
    --pred_len 1 \
    --learning_rate 0.001 \
    --loss 'SMAPE'
done

# python run.py \
#     --task_name short_term_forecast \
#     --is_training 0 \
#     --model_id poly_v1 \
#     --model TimesNet \
#     --data poly \
#     --root_path ./dataset/poly/ \
#     --seq_len 16 \
#     --label_len 1 \
#     --pred_len 1 \
#     --enc_in 1 \
#     --dec_in 1 \
#     --c_out 1 \
#     --features S \
#     --e_layers 2 \
#     --d_layers 1 \
#     --d_model 64 \
#     --d_ff 64 \
#     --top_k 3 \
#     --n_heads 4 \
#     --des poly_experiment \
#     --gpu 0