Final Train Runs

_parameters.ini

@@ -15,6 +15,9 @@ sr = 16000
 hop_length = 128
 n_fft = 256
 
+sample_segment_len=50
+sample_hop_len=20
+
 random_apply_chance = 0.7
 loudness_ratio = 0.0
 shift_ratio = 0.3
@@ -27,7 +30,7 @@ activation = gelu
 use_bias = True
 use_norm = True
 use_residual = True
-dropout = 0.2
+dropout = 0.21
 
 lat_dim = 32
 patch_size = 8

main.py

@@ -37,8 +37,6 @@ def run_lightning_loop(h_params, data_class, model_class, seed=69, additional_ca
         # Learning Rate Logger
         lr_logger = LearningRateMonitor(logging_interval='epoch')
 
-
-
         # Track best scores
         score_callback = BestScoresCallback(['PL_recall_score'])
 

multi_run.py

@@ -13,13 +13,13 @@ if __name__ == '__main__':
     hparams_dict = dict(seed=range(10),
                         model_name=['VisualTransformer'],
                         batch_size=[50],
-                        max_epochs=[250],
+                        max_epochs=[200],
                         random_apply_chance=[0.3],  # trial.suggest_float('random_apply_chance', 0.1, 0.5, step=0.1),
                         loudness_ratio=[0],  # trial.suggest_float('loudness_ratio', 0.0, 0.5, step=0.1),
                         shift_ratio=[0.3],  # trial.suggest_float('shift_ratio', 0.0, 0.5, step=0.1),
                         noise_ratio=[0.3],  # trial.suggest_float('noise_ratio', 0.0, 0.5, step=0.1),
                         mask_ratio=[0.3],  # trial.suggest_float('mask_ratio', 0.0, 0.5, step=0.1),
-                        lr=[5e-3],  # trial.suggest_uniform('lr', 1e-3, 3e-3),
+                        lr=[2e-3],  # trial.suggest_uniform('lr', 1e-3, 3e-3),
                         dropout=[0.2],  # trial.suggest_float('dropout', 0.0, 0.3, step=0.05),
                         lat_dim=[32],  # 2 ** trial.suggest_int('lat_dim', 1, 5, step=1),
                         mlp_dim=[16],  # 2 ** trial.suggest_int('mlp_dim', 1, 5, step=1),
@@ -28,7 +28,7 @@ if __name__ == '__main__':
                         attn_depth=[10],  # trial.suggest_int('attn_depth', 2, 14, step=4),
                         heads=[6],  # trial.suggest_int('heads', 2, 16, step=2),
                         scheduler=['CosineAnnealingWarmRestarts'],  # trial.suggest_categorical('scheduler', [None, 'LambdaLR']),
-                        lr_scheduler_parameter=[25],  # [0.98],
+                        lr_scheduler_parameter=[5],  # [0.98],
                         embedding_size=[30],  # trial.suggest_int('embedding_size', 12, 64, step=12),
                         loss=['ce_loss'],
                         sampler=['WeightedRandomSampler'],

util/optimizer_mixin.py

@@ -26,7 +26,7 @@ class OptimizerMixin:
         optimizer_dict.update(optimizer=optimizer)
 
         if self.params.scheduler == CosineAnnealingWarmRestarts.__name__:
-            scheduler = CosineAnnealingWarmRestarts(optimizer, self.params.lr_scheduler_parameter)
+            scheduler = CosineAnnealingWarmRestarts(optimizer, T_0=self.params.lr_scheduler_parameter)
         elif self.params.scheduler == LambdaLR.__name__:
             lr_reduce_ratio = self.params.lr_scheduler_parameter
             scheduler = LambdaLR(optimizer, lr_lambda=lambda epoch: lr_reduce_ratio ** epoch)