Upload autocap-full.yaml with huggingface_hub

autocap-full.yaml

@@ -47,6 +47,7 @@ model:
   use_clap_embeds: true
   meta_input: true
   add_special_tokens: True # If not then the meat data will start with Title:, Caption:, etc
+  resize_token_embeds: True
   meta_keys: ['video_caption', 'title']
   # meta_keys: ['video_caption', 'videollama_caption', 'title', 'description', 'subtitle', 'labels'] 
 
@@ -100,8 +101,8 @@ data_args:
   data: 
     metadata_root: "../dataset_preperation/data/metadata/dataset_root.json"
     train: ['32k_captioned_audiocaps', 'caption_audioset', 'wavcaps_audioset_strong', 'wavcaps_bbcsound', 'wavcaps_freesound', 'wavcaps_soundbible', 'clotho', 'fsd50k']
-    val: ['32k_captioned_audiocaps']
-    test: ['32k_captioned_audiocaps']
+    val: ['autocap']
+    test: ['autocap']
 
     keys_synonyms:
       gt_audio_caption:
@@ -110,6 +111,7 @@ data_args:
         - gt_caption
         - caption
         - gt_audio_caption
+        - autocap_caption
         - wavcaps_caption
       tags:
         - keywords