Create configuration_hubert_kmeans.py

configuration_hubert_kmeans.py

@@ -0,0 +1,25 @@
+"""HuBERT K-means configuration"""
+
+from transformers import PretrainedConfig
+
+
+class HubertKmeansConfig(PretrainedConfig):
+    """Configuration class for HuBERT K-means Quantizer."""
+    
+    model_type = "hubert_kmeans"
+    
+    def __init__(
+        self,
+        hubert_model_name: str = "facebook/hubert-base-ls960",
+        n_clusters: int = 200,
+        layer: int = 5,  # Which HuBERT layer to use (0-indexed, so 5 = 6th layer)
+        sample_rate: int = 16000,
+        freeze_hubert: bool = True,
+        **kwargs
+    ):
+        super().__init__(**kwargs)
+        self.hubert_model_name = hubert_model_name
+        self.n_clusters = n_clusters
+        self.layer = layer
+        self.sample_rate = sample_rate
+        self.freeze_hubert = freeze_hubert