fix: Change w2v2 word delimiter token to space

alexandrainst · Sep 20, 2023 · c329a88 · c329a88
1 parent 49c8063
commit c329a88
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/src/coral_models/wav2vec2.py b/src/coral_models/wav2vec2.py
@@ -115,7 +115,7 @@ def load_processor(self) -> Wav2Vec2Processor:
             pad_token="<pad>",
             bos_token="<s>",
             eos_token="</s>",
-            word_delimiter_token="|",
+            word_delimiter_token=" ",
         )
 
         # Set the `model_max_length` attribute of the tokenizer, if it hasn't been set,