[FIX] Unify API

.github/workflows/ci.yaml

@@ -40,4 +40,4 @@ jobs:
           uv pip install --system "numpy<2" ".[dev]"
 
       - name: Tests
-        run: nbdev_test --do_print --timing --n_workers 0 --flags polars
+        run: nbdev_test --do_print --timing --n_workers 0 --flags polars

action_files/test_models/src/evaluation.py

@@ -41,9 +41,12 @@ def evaluate(model: str, dataset: str, group: str):
 
 if __name__ == '__main__':
     groups = ['Monthly']
-    models = ['AutoDilatedRNN', 'RNN', 'TCN', 'DeepAR',
+    models = ['AutoDilatedRNN', 'RNN', 
+              'TCN', 
+              'DeepAR',
               'NHITS', 'TFT', 'AutoMLP', 'DLinear', 'VanillaTransformer',
-              'BiTCN', 'TiDE', 'DeepNPTS', 'NBEATS', 'KAN']
+              'BiTCN', 'TiDE', 'DeepNPTS', 'NBEATS', 'KAN'
+              ]
     datasets = ['M3']
     evaluation = [evaluate(model, dataset, group) for model, group in product(models, groups) for dataset in datasets]
     evaluation = [eval_ for eval_ in evaluation if eval_ is not None]

action_files/test_models/src/models.py

@@ -61,21 +61,22 @@ def main(dataset: str = 'M3', group: str = 'Monthly') -> None:
         "random_seed": tune.choice([1, 2, 3, 4, 5, 6, 7, 8, 9, 10]),
     }
     config_drnn = {'input_size': tune.choice([2 * horizon]),
-                   'encoder_hidden_size': tune.choice([124]),
+                   'encoder_hidden_size': tune.choice([16]),
                    "max_steps": 300,
                    "val_check_steps": 100,
-                   "random_seed": tune.choice([1, 2, 3, 4, 5, 6, 7, 8, 9, 10]),}
+                   "random_seed": tune.choice([1, 2, 3, 4, 5, 6, 7, 8, 9, 10]),
+                   "scaler_type": "minmax1"}
 
     models = [
         AutoDilatedRNN(h=horizon, loss=MAE(), config=config_drnn, num_samples=2, cpus=1),
-        RNN(h=horizon, input_size=2 * horizon, encoder_hidden_size=50, max_steps=300),
-        TCN(h=horizon, input_size=2 * horizon, encoder_hidden_size=20, max_steps=300),
+        RNN(h=horizon, input_size=2 * horizon, encoder_hidden_size=64, max_steps=300),
+        TCN(h=horizon, input_size=2 * horizon, encoder_hidden_size=64, max_steps=300),
         NHITS(h=horizon, input_size=2 * horizon, dropout_prob_theta=0.5, loss=MAE(), max_steps=1000, val_check_steps=500),
         AutoMLP(h=horizon, loss=MAE(), config=config, num_samples=2, cpus=1),
         DLinear(h=horizon, input_size=2 * horizon, loss=MAE(), max_steps=2000, val_check_steps=500),
         TFT(h=horizon, input_size=2 * horizon, loss=SMAPE(), hidden_size=64, scaler_type='robust', windows_batch_size=512, max_steps=1500, val_check_steps=500),
         VanillaTransformer(h=horizon, input_size=2 * horizon, loss=MAE(), hidden_size=64, scaler_type='minmax1', windows_batch_size=512, max_steps=1500, val_check_steps=500),
-        DeepAR(h=horizon, input_size=2 * horizon, scaler_type='minmax1', max_steps=1000),
+        DeepAR(h=horizon, input_size=2 * horizon, scaler_type='minmax1', max_steps=500),
         BiTCN(h=horizon, input_size=2 * horizon, loss=MAE(), dropout=0.0, max_steps=1000, val_check_steps=500),
         TiDE(h=horizon, input_size=2 * horizon, loss=MAE(), max_steps=1000, val_check_steps=500),
         DeepNPTS(h=horizon, input_size=2 * horizon, loss=MAE(), max_steps=1000, val_check_steps=500),

action_files/test_models/src/models2.py

@@ -2,35 +2,39 @@
 import time
 
 import fire
-import numpy as np
+# import numpy as np
 import pandas as pd
-import pytorch_lightning as pl
-import torch
+# import pytorch_lightning as pl
+# import torch
 
-import neuralforecast
+# import neuralforecast
 from neuralforecast.core import NeuralForecast
 
 from neuralforecast.models.gru import GRU
-from neuralforecast.models.rnn import RNN
-from neuralforecast.models.tcn import TCN
+# from neuralforecast.models.rnn import RNN
+# from neuralforecast.models.tcn import TCN
 from neuralforecast.models.lstm import LSTM
 from neuralforecast.models.dilated_rnn import DilatedRNN
-from neuralforecast.models.deepar import DeepAR
-from neuralforecast.models.mlp import MLP
-from neuralforecast.models.nhits import NHITS
-from neuralforecast.models.nbeats import NBEATS
+# from neuralforecast.models.deepar import DeepAR
+# from neuralforecast.models.mlp import MLP
+# from neuralforecast.models.nhits import NHITS
+# from neuralforecast.models.nbeats import NBEATS
 from neuralforecast.models.nbeatsx import NBEATSx
-from neuralforecast.models.tft import TFT
-from neuralforecast.models.vanillatransformer import VanillaTransformer
-from neuralforecast.models.informer import Informer
-from neuralforecast.models.autoformer import Autoformer
-from neuralforecast.models.patchtst import PatchTST
+# from neuralforecast.models.tft import TFT
+# from neuralforecast.models.vanillatransformer import VanillaTransformer
+# from neuralforecast.models.informer import Informer
+# from neuralforecast.models.autoformer import Autoformer
+# from neuralforecast.models.patchtst import PatchTST
 
 from neuralforecast.auto import (
-    AutoMLP, AutoNHITS, AutoNBEATS, AutoDilatedRNN, AutoTFT
+    # AutoMLP, 
+    AutoNHITS, 
+    AutoNBEATS, 
+    # AutoDilatedRNN, 
+    # AutoTFT
 )
 
-from neuralforecast.losses.pytorch import SMAPE, MAE
+from neuralforecast.losses.pytorch import MAE
 from ray import tune
 
 from src.data import get_data
@@ -49,32 +53,18 @@ def main(dataset: str = 'M3', group: str = 'Monthly') -> None:
         "scaler_type": "minmax1",
         "random_seed": tune.choice([1, 2, 3, 4, 5, 6, 7, 8, 9, 10]),
     }
-    config = {
-        "hidden_size": tune.choice([256, 512]),
-        "num_layers": tune.choice([2, 4]),
-        "input_size": tune.choice([2 * horizon]),
-        "max_steps": 1000,
-        "val_check_steps": 300,
-        "scaler_type": "minmax1",
-        "random_seed": tune.choice([1, 2, 3, 4, 5, 6, 7, 8, 9, 10]),
-    }
-    config_drnn = {'input_size': tune.choice([2 * horizon]),
-                   'encoder_hidden_size': tune.choice([124]),
-                   "max_steps": 300,
-                   "val_check_steps": 100,
-                   "random_seed": tune.choice([1, 2, 3, 4, 5, 6, 7, 8, 9, 10]),}
     models = [
-        LSTM(h=horizon, input_size=2 * horizon, encoder_hidden_size=50, max_steps=300),
-        DilatedRNN(h=horizon, input_size=2 * horizon, encoder_hidden_size=50, max_steps=300),
-        GRU(h=horizon, input_size=2 * horizon, encoder_hidden_size=50, max_steps=300),
+        LSTM(h=horizon, input_size=2 * horizon, encoder_hidden_size=64, max_steps=300),
+        DilatedRNN(h=horizon, input_size=2 * horizon, encoder_hidden_size=64, max_steps=300),
+        GRU(h=horizon, input_size=2 * horizon, encoder_hidden_size=64, max_steps=300),
         AutoNBEATS(h=horizon, loss=MAE(), config=config_nbeats, num_samples=2, cpus=1),
         AutoNHITS(h=horizon, loss=MAE(), config=config_nbeats, num_samples=2, cpus=1),
         NBEATSx(h=horizon, input_size=2 * horizon, loss=MAE(), max_steps=1000),
-        PatchTST(h=horizon, input_size=2 * horizon, patch_len=4, stride=4, loss=MAE(), scaler_type='minmax1', windows_batch_size=512, max_steps=1000, val_check_steps=500),
+        # PatchTST(h=horizon, input_size=2 * horizon, patch_len=4, stride=4, loss=MAE(), scaler_type='minmax1', windows_batch_size=512, max_steps=1000, val_check_steps=500),
     ]
 
     # Models
-    for model in models[:-1]:
+    for model in models:
         model_name = type(model).__name__
         print(50*'-', model_name, 50*'-')
         start = time.time()

action_files/test_models/src/multivariate_models.py

@@ -10,7 +10,7 @@
 from neuralforecast.models.tsmixer import TSMixer
 from neuralforecast.models.tsmixerx import TSMixerx
 from neuralforecast.models.itransformer import iTransformer
-# from neuralforecast.models.stemgnn import StemGNN
+# # from neuralforecast.models.stemgnn import StemGNN
 from neuralforecast.models.mlpmultivariate import MLPMultivariate
 from neuralforecast.models.timemixer import TimeMixer
 
@@ -26,13 +26,13 @@ def main(dataset: str = 'multivariate', group: str = 'ETTm2') -> None:
     train['ds'] = pd.to_datetime(train['ds'])
 
     models = [
-        SOFTS(h=horizon, n_series=7, input_size=2 * horizon, loss=MAE(), dropout=0.0, max_steps=1000, val_check_steps=500),
-        TSMixer(h=horizon, n_series=7, input_size=2 * horizon, loss=MAE(), dropout=0.0, max_steps=1000, val_check_steps=500),
-        TSMixerx(h=horizon, n_series=7, input_size=2*horizon, loss=MAE(), dropout=0.0, max_steps=1000, val_check_steps=500),
-        iTransformer(h=horizon, n_series=7, input_size=2 * horizon, loss=MAE(), dropout=0.0, max_steps=1000, val_check_steps=500),
-        # StemGNN(h=horizon, n_series=7, input_size=2*horizon, loss=MAE(), dropout_rate=0.0, max_steps=1000, val_check_steps=500),
-        MLPMultivariate(h=horizon, n_series=7, input_size=2*horizon, loss=MAE(), max_steps=1000, val_check_steps=500),
-        TimeMixer(h=horizon, n_series=7, input_size=2*horizon, loss=MAE(), dropout=0.0, max_steps=1000, val_check_steps=500)
+        SOFTS(h=horizon, n_series=7, input_size=2 * horizon, loss=MAE(), dropout=0.0, max_steps=500, val_check_steps=100, windows_batch_size=64, inference_windows_batch_size=64),
+        TSMixer(h=horizon, n_series=7, input_size=2 * horizon, loss=MAE(), dropout=0.0, max_steps=1000, val_check_steps=100, windows_batch_size=64, inference_windows_batch_size=64),
+        TSMixerx(h=horizon, n_series=7, input_size=2*horizon, loss=MAE(), dropout=0.0, max_steps=1000, val_check_steps=100, windows_batch_size=64, inference_windows_batch_size=64),
+        iTransformer(h=horizon, n_series=7, input_size=2 * horizon, loss=MAE(), dropout=0.0, max_steps=500, val_check_steps=100, windows_batch_size=64, inference_windows_batch_size=64),
+        # StemGNN(h=horizon, n_series=7, input_size=2*horizon, loss=MAE(), dropout_rate=0.0, max_steps=1000, val_check_steps=500, windows_batch_size=64, inference_windows_batch_size=64),
+        MLPMultivariate(h=horizon, n_series=7, input_size=2*horizon, loss=MAE(), max_steps=1000, val_check_steps=100, windows_batch_size=64, inference_windows_batch_size=64),
+        TimeMixer(h=horizon, n_series=7, input_size=2*horizon, loss=MAE(), dropout=0.0, max_steps=500, val_check_steps=100, windows_batch_size=64, inference_windows_batch_size=64)
     ]
 
     # Models

nbs/common.base_auto.ipynb

@@ -238,7 +238,11 @@
     "        self.callbacks = callbacks\n",
     "\n",
     "        # Base Class attributes\n",
-    "        self.SAMPLING_TYPE = cls_model.SAMPLING_TYPE\n",
+    "        self.EXOGENOUS_FUTR = cls_model.EXOGENOUS_FUTR\n",
+    "        self.EXOGENOUS_HIST = cls_model.EXOGENOUS_HIST\n",
+    "        self.EXOGENOUS_STAT = cls_model.EXOGENOUS_STAT\n",
+    "        self.MULTIVARIATE = cls_model.MULTIVARIATE    \n",
+    "        self.RECURRENT = cls_model.RECURRENT    \n",
     "\n",
     "    def __repr__(self):\n",
     "        return type(self).__name__ if self.alias is None else self.alias\n",