| { | |
| "vision": { | |
| "image_size": 224, | |
| "patch_size": 14, | |
| "hidden_size": 768, | |
| "num_hidden_layers": 12, | |
| "num_attention_heads": 12, | |
| "intermediate_size": 3072, | |
| "hidden_act": "gelu", | |
| "layer_norm_eps": 1e-06, | |
| "attention_dropout": 0.0, | |
| "image_channels": 3 | |
| }, | |
| "text": { | |
| "vocab_size": 50257, | |
| "hidden_size": 768, | |
| "intermediate_size": 3072, | |
| "num_hidden_layers": 12, | |
| "num_attention_heads": 12, | |
| "num_key_value_heads": 4, | |
| "max_position_embeddings": 512, | |
| "rms_norm_eps": 1e-05, | |
| "rope_theta": 10000.0, | |
| "tie_word_embeddings": true, | |
| "pad_token_id": 50256, | |
| "bos_token_id": 50256, | |
| "eos_token_id": 50256 | |
| }, | |
| "projection": { | |
| "vision_hidden_size": 768, | |
| "text_hidden_size": 768, | |
| "projection_size": 768, | |
| "num_layers": 2, | |
| "activation": "gelu", | |
| "dropout": 0.1 | |
| }, | |
| "image_token_id": 50257, | |
| "pad_token_id": 50256, | |
| "bos_token_id": 50256, | |
| "eos_token_id": 50256, | |
| "model_type": "multimodel-small" | |
| } | |