{ "_name_or_path": "google/vit-large-patch16-224-in21k", "architectures": [ "ViTForImageClassification" ], "attention_probs_dropout_prob": 0.0, "encoder_stride": 16, "hidden_act": "gelu", "hidden_dropout_prob": 0.0, "hidden_size": 1024, "id2label": { "0": "ain", "1": "al", "10": "ha", "11": "haa", "12": "jeem", "13": "kaaf", "14": "khaa", "15": "la", "16": "laam", "17": "meem", "18": "nun", "19": "ra", "2": "aleff", "20": "saad", "21": "seen", "22": "sheen", "23": "ta", "24": "taa", "25": "thaa", "26": "thal", "27": "toot", "28": "waw", "29": "ya", "3": "bb", "30": "yaa", "31": "zay", "4": "dal", "5": "dha", "6": "dhad", "7": "fa", "8": "gaaf", "9": "ghain" }, "image_size": 224, "initializer_range": 0.02, "intermediate_size": 4096, "label2id": { "ain": "0", "al": "1", "aleff": "2", "bb": "3", "dal": "4", "dha": "5", "dhad": "6", "fa": "7", "gaaf": "8", "ghain": "9", "ha": "10", "haa": "11", "jeem": "12", "kaaf": "13", "khaa": "14", "la": "15", "laam": "16", "meem": "17", "nun": "18", "ra": "19", "saad": "20", "seen": "21", "sheen": "22", "ta": "23", "taa": "24", "thaa": "25", "thal": "26", "toot": "27", "waw": "28", "ya": "29", "yaa": "30", "zay": "31" }, "layer_norm_eps": 1e-12, "model_type": "vit", "num_attention_heads": 16, "num_channels": 3, "num_hidden_layers": 24, "patch_size": 16, "problem_type": "single_label_classification", "qkv_bias": true, "torch_dtype": "float32", "transformers_version": "4.30.2" }