CONFIF_PATH = __file__
TRAIN_DIR = "/home/sds/hoanglv/Projects/TokenClassification_invoice/DATA/train"
TEST_DIR = "/home/sds/hoanglv/Projects/TokenClassification_invoice/DATA/test"
TOKENIZER_DIR = "Kie_Hoanglv/model/layoutxlm-base-tokenizer"
TOKENIZER_NAME = "microsoft/layoutxlm-base"
MODEL_WEIGHT = "microsoft/layoutxlm-base"
# pretrained model hyperparameter
MAX_SEQ_LENGTH = 512
IMG_SIZE = 224  # default

VN_list_char = "aAàÀảẢãÃáÁạẠăĂằẰẳẲẵẴắẮặẶâÂầẦẩẨẫẪấẤậẬbBcCdDđĐeEèÈẻẺẽẼéÉẹẸêÊềỀểỂễỄếẾệỆfFgGhHiIìÌỉỈĩĨíÍịỊjJkKlLmMnNoOòÒỏỎõÕóÓọỌôÔồỒổỔỗỖốỐộỘơƠờỜởỞỡỠớỚợỢpPqQrRsStTuUùÙủỦũŨúÚụỤưƯừỪửỬữỮứỨựỰvVwWxXyYỳỲỷỶỹỸýÝỵỴzZ0123456789!#$%&()*+,-./:;<=>?@[\]^_`{|}~"

DEVICE = "cuda:0"
SAVE_DIR = "runs/layoutxlm-base-17-10-2022-maxwords150_samplingv2"
BATCH_SIZE = 8
NUM_WORKER = 0
EPOCHS = 100
SAVE_INTERVAL = 1000
LR_RATE = 5e-6  # ori: 5e-5

# infer
MAX_N_WORDS = 150
TRAINED_DIR = "Kie_Hoanglv/model/layoutxlm-base-17-10-2022-maxwords150_samplingv2/last"
PRED_DIR = "/home/sds/hoanglv/Projects/TokenClassification_invoice/runs/infer/kie_e2e_pred_17-10-2022-maxwords150_samplingv2_rm_dup_boxes_test"
VISUALIZE_DIR = PRED_DIR + "/visualize"

KIE_LABELS = [
    # id invoice
    "no_key",
    "no_value",
    "form_key",
    "form_value",
    "serial_key",
    "serial_value",
    "date",
    # seller info
    "seller_company_name_key",
    "seller_company_name_value",
    "seller_tax_code_key",
    "seller_tax_code_value",
    "seller_address_value",
    "seller_address_key",
    "seller_mobile_key",
    "seller_mobile_value",
    # buyer info
    "buyer_name_key",
    "buyer_name_value",
    "buyer_company_name_value",
    "buyer_company_name_key",
    "buyer_tax_code_key",
    "buyer_tax_code_value",
    "buyer_address_key",
    "buyer_address_value",
    "buyer_mobile_key",
    "buyer_mobile_value",
    # money info
    "VAT_amount_key",
    "VAT_amount_value",
    "total_key",
    "total_value",
    "total_in_words_key",
    "total_in_words_value",
    "other",
]


SKIP_LABEL_EVAL = ["buyer_mobile_value"]