{ "_name_or_path": "microsoft/cvt-13", "architectures": [ "CvtForImageClassification" ], "attention_drop_rate": [ 0.0, 0.0, 0.0 ], "cls_token": [ false, false, true ], "depth": [ 1, 2, 10 ], "drop_path_rate": [ 0.0, 0.0, 0.1 ], "drop_rate": [ 0.0, 0.0, 0.0 ], "embed_dim": [ 64, 192, 384 ], "id2label": { "0": "Audemars Piguet", "1": "Breguet", "2": "Breitling", "3": "Maurice Lacroix", "4": "Omega", "5": "Patek Philippe", "6": "Rado", "7": "Rolex", "8": "Tissot", "9": "Vacheron Constantin" }, "image_size": 224, "initializer_range": 0.02, "kernel_qkv": [ 3, 3, 3 ], "label2id": { "Audemars Piguet": 0, "Breguet": 1, "Breitling": 2, "Maurice Lacroix": 3, "Omega": 4, "Patek Philippe": 5, "Rado": 6, "Rolex": 7, "Tissot": 8, "Vacheron Constantin": 9 }, "layer_norm_eps": 1e-12, "mlp_ratio": [ 4.0, 4.0, 4.0 ], "model_type": "cvt", "num_channels": 3, "num_heads": [ 1, 3, 6 ], "num_stages": 3, "padding_kv": [ 1, 1, 1 ], "padding_q": [ 1, 1, 1 ], "patch_padding": [ 2, 1, 1 ], "patch_sizes": [ 7, 3, 3 ], "patch_stride": [ 4, 2, 2 ], "pos_embed": [ false, false, false ], "problem_type": "single_label_classification", "qkv_bias": [ true, true, true ], "qkv_projection_method": [ "dw_bn", "dw_bn", "dw_bn" ], "stride_kv": [ 2, 2, 2 ], "stride_q": [ 1, 1, 1 ], "torch_dtype": "float32", "transformers_version": "4.40.0" }