| { | |
| "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, | |
| "num_attention_heads": 6, | |
| "num_hidden_layers": 6, | |
| "intermediate_size": 1536, | |
| "output_dim": 768, | |
| "hidden_dropout_prob": 0.0, | |
| "pad_token_id": 0, | |
| "tokenizer_class": "BertTokenizer", | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.48.0", | |
| "bank_enabled": true, | |
| "bank_n_experts": 5, | |
| "bank_n_anchors": 512, | |
| "bank_dim": 128, | |
| "bank_cv_target": 0.082, | |
| "consensus_models": [ | |
| "google-bert/bert-base-uncased", | |
| "answerdotai/ModernBERT-base", | |
| "FacebookAI/roberta-base", | |
| "albert/albert-base-v2", | |
| "distilbert/distilbert-base-uncased" | |
| ], | |
| "consensus_alignment": "generalized_procrustes_analysis", | |
| "consensus_equidistance": 0.0079, | |
| "training_data": "CaptionEmporium/conceptual-captions-cc12m-llavanext", | |
| "training_samples": 500000 | |
| } |