parent_command:forecast
usage: trans [--d-model D_MODEL] [--nhead NHEAD] [--num_encoder_layers NUM_ENCODER_LAYERS] [--num_decoder_layers NUM_DECODER_LAYERS] [--dim_feedforward DIM_FEEDFORWARD] [--activation {relu,gelu}] [--past-covariates PAST_COVARIATES]
             [--all-past-covariates] [--naive] [-d {AAPL}] [-c TARGET_COLUMN] [-n N_DAYS] [-t TRAIN_SPLIT] [-i INPUT_CHUNK_LENGTH] [-o OUTPUT_CHUNK_LENGTH] [--force-reset FORCE_RESET] [--save-checkpoints SAVE_CHECKPOINTS]
             [--model-save-name MODEL_SAVE_NAME] [--n-epochs N_EPOCHS] [--dropout DROPOUT] [--batch-size BATCH_SIZE] [--end S_END_DATE] [--start S_START_DATE] [--residuals] [--forecast-only] [--export-pred-raw]
             [--metric {rmse,mse,mape,smape}]

Perform Transformer Forecast: https://unit8co.github.io/darts/generated_api/darts.models.forecasting.transformer_model.html

optional arguments:
  --d-model D_MODEL     Number of expected features in inputs. (default: 64)
  --nhead NHEAD         Number of head in the attention mechanism. (default: 4)
  --num_encoder_layers NUM_ENCODER_LAYERS
                        The number of encoder layers in the encoder. (default: 3)
  --num_decoder_layers NUM_DECODER_LAYERS
                        The number of decoder layers in the encoder. (default: 3)
  --dim_feedforward DIM_FEEDFORWARD
                        The dimension of the feedforward model. (default: 512)
  --activation {relu,gelu}
                        Number of LSTM layers. (default: relu)
  --past-covariates PAST_COVARIATES
                        Past covariates(columns/features) in same dataset. Comma separated. (default: None)
  --all-past-covariates
                        Adds all rows as past covariates except for date and the target column. (default: False)
  --naive               Show the naive baseline for a model. (default: False)
  -d {AAPL}, --dataset {AAPL}
                        The name of the dataset you want to select (default: None)
  -c TARGET_COLUMN, --target-column TARGET_COLUMN
                        The name of the specific column you want to use (default: close)
  -n N_DAYS, --n-days N_DAYS
                        prediction days. (default: 5)
  -t TRAIN_SPLIT, --train-split TRAIN_SPLIT
                        Start point for rolling training and forecast window. 0.0-1.0 (default: 0.85)
  -i INPUT_CHUNK_LENGTH, --input-chunk-length INPUT_CHUNK_LENGTH
                        Number of past time steps for forecasting module at prediction time. (default: 14)
  -o OUTPUT_CHUNK_LENGTH, --output-chunk-length OUTPUT_CHUNK_LENGTH
                        The length of the forecast of the model. (default: 5)
  --force-reset FORCE_RESET
                        If set to True, any previously-existing model with the same name will be reset (all checkpoints will be discarded). (default: True)
  --save-checkpoints SAVE_CHECKPOINTS
                        Whether to automatically save the untrained model and checkpoints. (default: True)
  --model-save-name MODEL_SAVE_NAME
                        Name of the model to save. (default: trans_model)
  --n-epochs N_EPOCHS   Number of epochs over which to train the model. (default: 300)
  --dropout DROPOUT     Fraction of neurons affected by Dropout, from 0 to 1. (default: 0)
  --batch-size BATCH_SIZE
                        Number of time series (input and output) used in each training pass (default: 32)
  --end S_END_DATE      The end date (format YYYY-MM-DD) to select for testing (default: None)
  --start S_START_DATE  The start date (format YYYY-MM-DD) to select for testing (default: None)
  --residuals           Show the residuals for the model. (default: False)
  --forecast-only       Do not plot the historical data without forecasts. (default: False)
  --export-pred-raw     Export predictions to a csv file. (default: False)
  --metric {rmse,mse,mape,smape}
                        Calculate precision based on a specific metric (rmse, mse, mape) (default: mape)


Examples:
- Perform a Transformer Forecast on the default dataset with default parameters: forecast/trans
- Forecast <SYMBOL> for 10 days using specific input and output chunk lengths: stocks/load <SYMBOL>/forecast/trans -n 10 -i 20 -o 10
- Train a Transformer Forecast on <COIN> with a custom train split and 7-day forecast: crypto/load <COIN>/forecast/trans -t 0.9 -n 7
- Use a specific target column and activation function for forecasting <FUTURE>: futures/load <FUTURE>/forecast/trans -c open --activation gelu
- Forecast <SYMBOL> with custom model parameters and save the model with a custom name: stocks/load <SYMBOL>/forecast/trans --d-model 128 --nhead 8 --model-save-name custom_trans_model
- Perform a Transformer Forecast on <PAIR> with specific start and end dates: forex/load <PAIR>/forecast/trans --start 2020-01-01 --end 2021-12-31
- Show residuals and use a specific metric for a Transformer Forecast on <SYMBOL>: stocks/load <SYMBOL>/forecast/trans --residuals --metric rmse
- Export raw predictions to a CSV file for a Transformer Forecast on <COIN>: crypto/load <COIN>/forecast/trans --export-pred-raw
- Forecast only, without historical data, for <FUTURE> using a Transformer Forecast: futures/load <FUTURE>/forecast/trans --forecast-only
- Perform a naive baseline comparison for a Transformer Forecast on <SYMBOL>: stocks/load <SYMBOL>/forecast/trans --naive