| {"text_encoder": {"model_type": "bert", "dim": 768, "context_dim": 384, "vocab_size": 30522, "padding_idx": 0, "num_layers": 4, "num_heads": 12, "embedding_dim": 256, "multimodal_layers_ids": [2, 3], "head_one_neuron": false, "pooling": "cls", "max_position_embeddings": 64, "dropout_prob": 0.1}, "image_encoder": {"dim": 384, "patch_size": 16, "image_size": 224, "num_layers": 12, "num_heads": 6, "embedding_dim": 256, "pooling": "cls"}} |