{ "_name_or_path": "microsoft/beit-base-patch16-224", "add_fpn": false, "architectures": [ "BeitForImageClassification" ], "attention_probs_dropout_prob": 0.0, "auxiliary_channels": 256, "auxiliary_concat_input": false, "auxiliary_loss_weight": 0.4, "auxiliary_num_convs": 1, "drop_path_rate": 0.1, "hidden_act": "gelu", "hidden_dropout_prob": 0.0, "hidden_size": 768, "id2label": { "0": "Italian", "1": "Russian", "2": "Mexican", "3": "French", "4": "Belgian", "5": "Spanish", "6": "Dutch", "7": "Austrian", "8": "Flemish", "9": "Spanish,Greek", "10": "German", "11": "French,British", "12": "French,Jewish,Belarusian", "13": "British", "14": "Norwegian", "15": "German,Swiss", "16": "American" }, "image_size": 224, "initializer_range": 0.02, "intermediate_size": 3072, "label2id": { "American": 16, "Austrian": 7, "Belgian": 4, "British": 13, "Dutch": 6, "Flemish": 8, "French": 3, "French,British": 11, "French,Jewish,Belarusian": 12, "German": 10, "German,Swiss": 15, "Italian": 0, "Mexican": 2, "Norwegian": 14, "Russian": 1, "Spanish": 5, "Spanish,Greek": 9 }, "layer_norm_eps": 1e-12, "layer_scale_init_value": 0.1, "model_type": "beit", "num_attention_heads": 12, "num_channels": 3, "num_hidden_layers": 12, "out_features": [ "stage12" ], "out_indices": [ 12 ], "patch_size": 16, "pool_scales": [ 1, 2, 3, 6 ], "problem_type": "single_label_classification", "reshape_hidden_states": true, "semantic_loss_ignore_index": 255, "stage_names": [ "stem", "stage1", "stage2", "stage3", "stage4", "stage5", "stage6", "stage7", "stage8", "stage9", "stage10", "stage11", "stage12" ], "torch_dtype": "float32", "transformers_version": "4.46.3", "use_absolute_position_embeddings": false, "use_auxiliary_head": true, "use_mask_token": false, "use_mean_pooling": true, "use_relative_position_bias": true, "use_shared_relative_position_bias": false, "vocab_size": 8192 }