Update users/berger

Author: Simon Berger

users/berger/network/helpers/label_context.py

@@ -263,6 +263,7 @@ def add_context_1_decoder(
 def add_context_1_decoder_recog(
     network: Dict,
     num_outputs: int,
+    blank_idx: int = 0,
     encoder: str = "encoder",
     embedding_size: int = 128,
     dec_mlp_args: Dict = {},
@@ -351,9 +352,21 @@ def add_context_1_decoder_recog(
             "reuse_params": "output",
         }
 
+        assert blank_idx == 0, "Blank idx != 0 not implemented for ilm"
+        # Set p(blank) = 1 and re-normalize the non-blank probs
+        # so we want P'[b, 0] = 1, sum(P'[b, 1:]) = 1, given a normalized tensor P, i.e. sum(P[b, :]) = 1
+        # in log space logP'[b, 0] = 0, sum(exp(logP'[b, 1:])) = 1
+        # so set logP'[b, 1:] <- logP[b, 1:] - log(1 - exp(P[b, 0]))
+        # then sum(exp(logP'[b, 1:])) = sum(P[1:] / (1 - exp(P[b, 0]))) = sum(P[b, 1:]) / sum(b, P[1:]) = 1
+        output_unit["ilm_renorm"] = {
+            "class": "eval",
+            "from": ["ilm"],
+            "eval": "tf.concat([tf.zeros(tf.shape(source(0)[:, :1])), source(0)[:, 1:] - tf.math.log(1.0 - tf.exp(source(0)[:, :1]))], axis=-1)",
+        }
+
         output_unit["output_sub_ilm"] = {
             "class": "eval",
-            "from": ["output", "ilm"],
+            "from": ["output", "ilm_renorm"],
             "eval": f"source(0) - {ilm_scale} * source(1)",
         }
 

users/berger/recipe/rasr/label_tree_and_scorer.py

@@ -78,7 +78,7 @@ def __init__(
         use_prior: bool = False,
         prior_scale: float = 0.6,
         prior_file: Optional[tk.Path] = None,
-        extra_args: Dict = {},
+        extra_args: Optional[Dict] = None,
     ):
         self.config = rasr.RasrConfig()
         self.post_config = rasr.RasrConfig()
@@ -102,13 +102,64 @@ def __init__(
             self.config.priori_scale = prior_scale
 
         # sprint key values #
-        for key, value in extra_args.items():
-            self.config[key.replace("_", "-")] = value
+        if extra_args is not None:
+            for key, value in extra_args.items():
+                self.config[key.replace("_", "-")] = value
 
     @property
     def scorer_type(self):
         return self.config.label_scorer_type
 
+    @property
+    def scale(self):
+        return self.config.scale
+
+    @property
+    def label_file(self):
+        if self.config._get("label-file") is not None:
+            return self.config.label_file
+        return None
+
+    @property
+    def num_classes(self):
+        if self.config._get("number-of-classes") is not None:
+            return self.config.number_of_classes
+        return None
+
+    @property
+    def use_prior(self):
+        if self.config._get("use-prior") is not None:
+            return self.config["use-prior"]
+        return False
+
+    @property
+    def prior_scale(self):
+        if self.config._get("priori-scale") is not None:
+            return self.config["priori-scale"]
+        return 1.0
+
+    @property
+    def prior_file(self):
+        if self.config._get("prior-file") is not None:
+            return self.config["prior-file"]
+        return None
+
+    @property
+    def extra_args(self):
+        return {
+            key: val
+            for key, val in self.config._items()
+            if key not in [
+                "label-scorer-type",
+                "scale",
+                "label-file",
+                "number-of-classes",
+                "use-prior",
+                "priori-scale",
+                "prior-file",
+            ]
+        }
+
     def apply_config(
         self,
         path: str,