|
{ |
|
"_name_or_path": "cahya/NusaBert-v1.3", |
|
"architectures": [ |
|
"ModernBertForTokenClassification" |
|
], |
|
"attention_bias": false, |
|
"attention_dropout": 0.0, |
|
"bos_token_id": 65003, |
|
"classifier_activation": "gelu", |
|
"classifier_bias": false, |
|
"classifier_dropout": 0.0, |
|
"classifier_pooling": "cls", |
|
"cls_token_id": 65003, |
|
"decoder_bias": true, |
|
"deterministic_flash_attn": false, |
|
"embedding_dropout": 0.0, |
|
"eos_token_id": 65001, |
|
"finetuning_task": "ner", |
|
"global_attn_every_n_layers": 3, |
|
"global_rope_theta": 160000.0, |
|
"hidden_activation": "gelu", |
|
"hidden_size": 768, |
|
"id2label": { |
|
"0": "B-CRD", |
|
"1": "B-DAT", |
|
"2": "B-EVT", |
|
"3": "B-FAC", |
|
"4": "B-GPE", |
|
"5": "B-LAN", |
|
"6": "B-LAW", |
|
"7": "B-LOC", |
|
"8": "B-MON", |
|
"9": "B-NOR", |
|
"10": "B-ORD", |
|
"11": "B-ORG", |
|
"12": "B-PER", |
|
"13": "B-PRC", |
|
"14": "B-PRD", |
|
"15": "B-QTY", |
|
"16": "B-REG", |
|
"17": "B-TIM", |
|
"18": "B-WOA", |
|
"19": "I-CRD", |
|
"20": "I-DAT", |
|
"21": "I-EVT", |
|
"22": "I-FAC", |
|
"23": "I-GPE", |
|
"24": "I-LAN", |
|
"25": "I-LAW", |
|
"26": "I-LOC", |
|
"27": "I-MON", |
|
"28": "I-NOR", |
|
"29": "I-ORD", |
|
"30": "I-ORG", |
|
"31": "I-PER", |
|
"32": "I-PRC", |
|
"33": "I-PRD", |
|
"34": "I-QTY", |
|
"35": "I-REG", |
|
"36": "I-TIM", |
|
"37": "I-WOA", |
|
"38": "O" |
|
}, |
|
"initializer_cutoff_factor": 2.0, |
|
"initializer_range": 0.02, |
|
"intermediate_size": 1152, |
|
"label2id": { |
|
"B-CRD": 0, |
|
"B-DAT": 1, |
|
"B-EVT": 2, |
|
"B-FAC": 3, |
|
"B-GPE": 4, |
|
"B-LAN": 5, |
|
"B-LAW": 6, |
|
"B-LOC": 7, |
|
"B-MON": 8, |
|
"B-NOR": 9, |
|
"B-ORD": 10, |
|
"B-ORG": 11, |
|
"B-PER": 12, |
|
"B-PRC": 13, |
|
"B-PRD": 14, |
|
"B-QTY": 15, |
|
"B-REG": 16, |
|
"B-TIM": 17, |
|
"B-WOA": 18, |
|
"I-CRD": 19, |
|
"I-DAT": 20, |
|
"I-EVT": 21, |
|
"I-FAC": 22, |
|
"I-GPE": 23, |
|
"I-LAN": 24, |
|
"I-LAW": 25, |
|
"I-LOC": 26, |
|
"I-MON": 27, |
|
"I-NOR": 28, |
|
"I-ORD": 29, |
|
"I-ORG": 30, |
|
"I-PER": 31, |
|
"I-PRC": 32, |
|
"I-PRD": 33, |
|
"I-QTY": 34, |
|
"I-REG": 35, |
|
"I-TIM": 36, |
|
"I-WOA": 37, |
|
"O": 38 |
|
}, |
|
"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": 65002, |
|
"reference_compile": false, |
|
"repad_logits_with_grad": false, |
|
"sep_token_id": 65001, |
|
"sparse_pred_ignore_index": -100, |
|
"sparse_prediction": false, |
|
"torch_dtype": "float32", |
|
"transformers_version": "4.49.0", |
|
"vocab_size": 65107 |
|
} |
|
|