ENH: Determine network kwargs dynamically in vak.models.get (#680)

* Use inspect to get network kwargs in vak.models.get

* Fix error message that used non-existent variable in vak.models.get

* Use registry.MODEL_FAMILY_FROM_NAME when determining net kwargs in vak.models.get

src/vak/models/get.py

@@ -1,6 +1,7 @@
 """Function that gets an instance of a model,
 given its name and a configuration as a dict."""
 from __future__ import annotations
+import inspect
 from typing import Callable
 
 from . import registry
@@ -52,20 +53,29 @@ def get(name: str,
             f"Valid model names are: {registry.MODEL_NAMES}"
         ) from e
 
-    # still need to special case model logic here
-    if name in ('TweetyNet', 'TeenyTweetyNet', 'ED_TCN'):
-        num_input_channels = input_shape[-3]
-        num_freqbins = input_shape[-2]
-        config["network"].update(
-            num_classes=num_classes,
-            num_input_channels=num_input_channels,
-            num_freqbins=num_freqbins
-        )
-    else:
-        model_names = list(all_models_dict.keys())
-        raise ValueError(
-            f"Invalid model name: '{name}'.\nValid model names are: {model_names}"
+    model_family = registry.MODEL_FAMILY_FROM_NAME[name]
+
+    if model_family == 'FrameClassificationModel':
+        # still need to special case model logic here
+        net_init_params = list(
+            inspect.signature(
+                model_class.definition.network.__init__
+            ).parameters.keys()
         )
+        if ('num_input_channels' in net_init_params) and ('num_freqbins' in net_init_params):
+            num_input_channels = input_shape[-3]
+            num_freqbins = input_shape[-2]
+            config["network"].update(
+                num_classes=num_classes,
+                num_input_channels=num_input_channels,
+                num_freqbins=num_freqbins
+            )
+        else:
+            raise ValueError(
+                f"Detected that model with name '{name}' was family '{model_family}', but "
+                f"unable to determine network init arguments for model. Currently all models "
+                f"in this family must have networks with parameters ``num_input_channels`` and ``num_freqbins``"
+            )
 
     model = model_class.from_config(config=config, labelmap=labelmap, post_tfm=post_tfm)