davidberenstein1957's picture
Training in progress, epoch 1
14a6760 verified
{
"_name_or_path": "answerdotai/ModernBERT-base",
"architectures": [
"ModernBertForSequenceClassification"
],
"attention_bias": false,
"attention_dropout": 0.0,
"bos_token_id": 50281,
"classifier_activation": "gelu",
"classifier_bias": false,
"classifier_dropout": 0.0,
"classifier_pooling": "mean",
"cls_token_id": 50281,
"decoder_bias": true,
"deterministic_flash_attn": false,
"embedding_dropout": 0.0,
"eos_token_id": 50282,
"global_attn_every_n_layers": 3,
"global_rope_theta": 160000.0,
"gradient_checkpointing": false,
"hidden_activation": "gelu",
"hidden_size": 768,
"id2label": {
"0": "business-and-industrial",
"1": "books-and-literature",
"10": "home-and-garden",
"11": "adult",
"12": "science",
"13": "food-and-drink",
"14": "real-estate",
"15": "news",
"16": "jobs-and-education",
"17": "health",
"18": "hobbies-and-leisure",
"19": "games",
"2": "beauty-and-fitness",
"20": "computers-and-electronics",
"21": "arts-and-entertainment",
"22": "travel-and-transportation",
"23": "finance",
"24": "law-and-government",
"25": "sensitive-subjects",
"3": "autos-and-vehicles",
"4": "people-and-society",
"5": "sports",
"6": "shopping",
"7": "online-communities",
"8": "pets-and-animals",
"9": "internet-and-telecom"
},
"initializer_cutoff_factor": 2.0,
"initializer_range": 0.02,
"intermediate_size": 1152,
"label2id": {
"adult": "11",
"arts-and-entertainment": "21",
"autos-and-vehicles": "3",
"beauty-and-fitness": "2",
"books-and-literature": "1",
"business-and-industrial": "0",
"computers-and-electronics": "20",
"finance": "23",
"food-and-drink": "13",
"games": "19",
"health": "17",
"hobbies-and-leisure": "18",
"home-and-garden": "10",
"internet-and-telecom": "9",
"jobs-and-education": "16",
"law-and-government": "24",
"news": "15",
"online-communities": "7",
"people-and-society": "4",
"pets-and-animals": "8",
"real-estate": "14",
"science": "12",
"sensitive-subjects": "25",
"shopping": "6",
"sports": "5",
"travel-and-transportation": "22"
},
"layer_norm_eps": 1e-05,
"local_attention": 128,
"local_rope_theta": 10000.0,
"max_position_embeddings": 8192,
"mlp_bias": false,
"mlp_dropout": 0.0,
"model_type": "modernbert",
"norm_bias": false,
"norm_eps": 1e-05,
"num_attention_heads": 12,
"num_hidden_layers": 22,
"pad_token_id": 50283,
"position_embedding_type": "absolute",
"problem_type": "single_label_classification",
"reference_compile": false,
"sep_token_id": 50282,
"sparse_pred_ignore_index": -100,
"sparse_prediction": false,
"torch_dtype": "float32",
"transformers_version": "4.48.0.dev0",
"vocab_size": 50368
}