longformer_pico_model / config.json
abhiman23897's picture
Training in progress, epoch 1
34c1014
{
"_name_or_path": "allenai/longformer-base-4096",
"architectures": [
"LongformerForTokenClassification"
],
"attention_mode": "longformer",
"attention_probs_dropout_prob": 0.1,
"attention_window": [
512,
512,
512,
512,
512,
512,
512,
512,
512,
512,
512,
512
],
"bos_token_id": 0,
"eos_token_id": 2,
"gradient_checkpointing": false,
"hidden_act": "gelu",
"hidden_dropout_prob": 0.1,
"hidden_size": 768,
"id2label": {
"0": "O",
"1": "B-total-participants",
"2": "I-total-participants",
"3": "B-intervention-participants",
"4": "I-intervention-participants",
"5": "B-control-participants",
"6": "I-control-participants",
"7": "B-age",
"8": "I-age",
"9": "B-eligibility",
"10": "I-eligibility",
"11": "B-ethinicity",
"12": "I-ethinicity",
"13": "B-condition",
"14": "I-condition",
"15": "B-location",
"16": "I-location",
"17": "B-intervention",
"18": "I-intervention",
"19": "B-control",
"20": "I-control",
"21": "B-outcome",
"22": "I-outcome",
"23": "B-outcome-measure",
"24": "I-outcome-measure",
"25": "B-iv-bin-abs",
"26": "I-iv-bin-abs",
"27": "B-cv-bin-abs",
"28": "I-cv-bin-abs",
"29": "B-iv-bin-percent",
"30": "I-iv-bin-percent",
"31": "B-cv-bin-percent",
"32": "I-cv-bin-percent",
"33": "B-iv-cont-mean",
"34": "I-iv-cont-mean",
"35": "B-cv-cont-mean",
"36": "I-cv-cont-mean",
"37": "B-iv-cont-median",
"38": "I-iv-cont-median",
"39": "B-cv-cont-median",
"40": "I-cv-cont-median",
"41": "B-iv-cont-sd",
"42": "I-iv-cont-sd",
"43": "B-cv-cont-sd",
"44": "I-cv-cont-sd",
"45": "B-iv-cont-q1",
"46": "I-iv-cont-q1",
"47": "B-cv-cont-q1",
"48": "I-cv-cont-q1",
"49": "B-iv-cont-q3",
"50": "I-iv-cont-q3",
"51": "B-cv-cont-q3",
"52": "I-cv-cont-q3"
},
"ignore_attention_mask": false,
"initializer_range": 0.02,
"intermediate_size": 3072,
"label2id": {
"B-age": 7,
"B-condition": 13,
"B-control": 19,
"B-control-participants": 5,
"B-cv-bin-abs": 27,
"B-cv-bin-percent": 31,
"B-cv-cont-mean": 35,
"B-cv-cont-median": 39,
"B-cv-cont-q1": 47,
"B-cv-cont-q3": 51,
"B-cv-cont-sd": 43,
"B-eligibility": 9,
"B-ethinicity": 11,
"B-intervention": 17,
"B-intervention-participants": 3,
"B-iv-bin-abs": 25,
"B-iv-bin-percent": 29,
"B-iv-cont-mean": 33,
"B-iv-cont-median": 37,
"B-iv-cont-q1": 45,
"B-iv-cont-q3": 49,
"B-iv-cont-sd": 41,
"B-location": 15,
"B-outcome": 21,
"B-outcome-measure": 23,
"B-total-participants": 1,
"I-age": 8,
"I-condition": 14,
"I-control": 20,
"I-control-participants": 6,
"I-cv-bin-abs": 28,
"I-cv-bin-percent": 32,
"I-cv-cont-mean": 36,
"I-cv-cont-median": 40,
"I-cv-cont-q1": 48,
"I-cv-cont-q3": 52,
"I-cv-cont-sd": 44,
"I-eligibility": 10,
"I-ethinicity": 12,
"I-intervention": 18,
"I-intervention-participants": 4,
"I-iv-bin-abs": 26,
"I-iv-bin-percent": 30,
"I-iv-cont-mean": 34,
"I-iv-cont-median": 38,
"I-iv-cont-q1": 46,
"I-iv-cont-q3": 50,
"I-iv-cont-sd": 42,
"I-location": 16,
"I-outcome": 22,
"I-outcome-measure": 24,
"I-total-participants": 2,
"O": 0
},
"layer_norm_eps": 1e-05,
"max_position_embeddings": 4098,
"model_type": "longformer",
"num_attention_heads": 12,
"num_hidden_layers": 12,
"onnx_export": false,
"pad_token_id": 1,
"sep_token_id": 2,
"torch_dtype": "float32",
"transformers_version": "4.30.2",
"type_vocab_size": 1,
"vocab_size": 50265
}