Update config.json

config.json

@@ -1,8 +1,10 @@
 {
-  "model_type": "caption_encoder",
-  "architectures": [
-    "CaptionEncoder"
-  ],
+  "model_type": "caption_bert",
+  "auto_map": {
+    "AutoConfig": "modeling_caption_bert.CaptionBertConfig",
+    "AutoModel": "modeling_caption_bert.CaptionBertModel"
+  },
+  "architectures": ["CaptionBertModel"],
   "vocab_size": 30522,
   "max_position_embeddings": 8192,
   "hidden_size": 384,
@@ -10,9 +12,11 @@
   "num_hidden_layers": 6,
   "intermediate_size": 1536,
   "output_dim": 768,
-  "hidden_dropout_prob": 0.1,
+  "hidden_dropout_prob": 0.0,
   "pad_token_id": 0,
-  "tokenizer": "google-bert/bert-base-uncased",
+  "tokenizer_class": "BertTokenizer",
+  "torch_dtype": "float32",
+  "transformers_version": "4.48.0",
   "consensus_models": [
     "google-bert/bert-base-uncased",
     "answerdotai/ModernBERT-base",
@@ -20,8 +24,8 @@
     "albert/albert-base-v2",
     "distilbert/distilbert-base-uncased"
   ],
-  "alignment": "whitened_procrustes",
-  "consensus": "normalized_centroid",
-  "cv_target": 0.084,
-  "torch_dtype": "float32"
+  "consensus_alignment": "whitened_procrustes",
+  "consensus_cv_target": 0.084,
+  "training_data": "CaptionEmporium/conceptual-captions-cc12m-llavanext",
+  "training_samples": 500000
 }