fix naming 'sliding_window'

data/collate/data_collate_factory.py

@@ -17,7 +17,7 @@ def create_data_collate(architecture, embeddings, resolution):
     """
     if architecture == "autoencoder":
         return AutoencoderCollate(embeddings)
-    if architecture in ("encoder_only", 'pytorch', 'fast', 'fast-recurrent', 'sliding-window'):
+    if architecture in ("encoder_only", 'pytorch', 'fast', 'fast-recurrent', 'fast_recurrent', 'sliding_window'):
         return EncoderOnlyCollate()
     if architecture == "encoder_decoder":
         return EncoderDecoderCollate(embeddings)

modules/architecture/architecture_factory.py

@@ -62,13 +62,13 @@ def create_architecture(
         return Transformer(**kwargs, num_decoders=len(token_embedding) - 1)
     elif architecture == "pytorch":
         return PytorchTransformer(**kwargs)
-    elif architecture == "sliding-window":
+    elif architecture == "sliding_window":
         return SlidingWindowTransformer(**kwargs)
     elif architecture == "fast":
         # include `pytorch-fast-transformers` as an optional module
         from .fast_transformer import FastTransformer
         return FastTransformer(**kwargs)
-    elif architecture == "fast-recurrent":
+    elif architecture in ("fast-recurrent", "fast_recurrent"):
         # include `pytorch-fast-transformers` as an optional module
         from .fast_recurrent_transformer import FastRecurrentTransformer
         return FastRecurrentTransformer(**kwargs)

sample/layer_sampler/recurrent_sampler.py

@@ -13,7 +13,7 @@ from ..token_generator.recurrent import create_recurrent_token_generator
 
 class RecurrentSampler:
     def __init__(self, model, head, spatial_dim, max_resolution, position_encoding, device, **_):
-        """ Provides a basic implementation of the sampler for the 'fast-recurrent-transformer' architecture.
+        """ Provides a basic implementation of the sampler for the 'fast_recurrent-transformer' architecture.
 
         Args:
             model: Model which is used for sampling.

sample/layer_sampler/sampler_factory.py

@@ -37,9 +37,9 @@ def create_sampler(
 
     if architecture == "autoencoder":
         return AutoencoderSampler(**kwargs)
-    elif architecture in ("encoder_only", "fast", "pytorch", "sliding-window"):
+    elif architecture in ("encoder_only", "fast", "pytorch", "sliding_window"):
         return EncoderOnlySampler(**kwargs)
-    elif architecture in ("fast-recurrent"):
+    elif architecture in ("fast-recurrent", "fast_recurrent"):
         return RecurrentSampler(**kwargs)
     elif architecture == "encoder_decoder":
         return EncoderDecoderSampler(**kwargs)

sample/shape_sampler.py

@@ -20,7 +20,7 @@ class ShapeSampler:
         pl_module = ShapeTransformer.load_from_checkpoint(checkpoint_path)
         if fast_recurrent is True and pl_module.hparams['architecture'] == 'fast':
             print("Reload model as a recurrent implementation for a major improvement of inference time.")
-            pl_module = ShapeTransformer.load_from_checkpoint(checkpoint_path, architecture='fast-recurrent')
+            pl_module = ShapeTransformer.load_from_checkpoint(checkpoint_path, architecture='fast_recurrent')
         pl_module.freeze()
 
         # extract hyperparameters from the model