Fix: Add missing CACHE_DIR to data.manager for training cache support

data/manager.py

@@ -8,12 +8,14 @@ DATA_DIR = os.path.abspath(DATA_DIR)
 os.makedirs(DATA_DIR, exist_ok=True)
 
 # Export BASE_DIR for use in other modules
-__all__ = ['DATA_DIR', 'AUDIO_DIR', 'ENTITIES_PATH', 'MODEL_OUTPUT_DIR', 'BASE_DIR']
+__all__ = ['DATA_DIR', 'AUDIO_DIR', 'ENTITIES_PATH', 'MODEL_OUTPUT_DIR', 'BASE_DIR', 'CACHE_DIR']
 
 AUDIO_DIR = os.path.join(DATA_DIR, "audio_files")
 os.makedirs(AUDIO_DIR, exist_ok=True)
 ENTITIES_PATH = os.path.join(DATA_DIR, "caribbean_entities.json")
 MODEL_OUTPUT_DIR = os.path.join(DATA_DIR, "owsm_caribbean_finetuned")
+CACHE_DIR = os.path.join(DATA_DIR, "processed_datasets_cache")
+os.makedirs(CACHE_DIR, exist_ok=True)
 
 # Debug paths
 print(f"DEBUG: DATA_DIR = {DATA_DIR}")