| { | |
| "architectures": [ | |
| "ViTMAEForPreTraining" | |
| ], | |
| "attention_probs_dropout_prob": 0.0, | |
| "decoder_hidden_size": 512, | |
| "decoder_intermediate_size": 2048, | |
| "decoder_num_attention_heads": 16, | |
| "decoder_num_hidden_layers": 8, | |
| "hidden_act": "gelu", | |
| "hidden_dropout_prob": 0.0, | |
| "hidden_size": 1024, | |
| "image_size": 224, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 4096, | |
| "layer_norm_eps": 1e-12, | |
| "mask_ratio": 0.75, | |
| "model_type": "vit_mae", | |
| "norm_pix_loss": false, | |
| "num_attention_heads": 16, | |
| "num_channels": 3, | |
| "num_hidden_layers": 24, | |
| "patch_size": 16, | |
| "qkv_bias": true, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.16.0.dev0" | |
| } | |