thak123's picture
Update config.py
70e3867
raw
history blame
997 Bytes
import transformers
import os
import torch
MAX_LEN = 150 #256
TRAIN_BATCH_SIZE = 8
VALID_BATCH_SIZE = 4
EPOCHS = 5
# Folder to contain all the datasets
from huggingface_hub import hf_hub_download
hf_hub_download(repo_id="thak123/bert-emoji-latvian-twitter-classifier", filename="model.bin",local_dir="./")
# import requests
# URL = "https://huggingface.co/thak123/bert-emoji-latvian-twitter-classifier/resolve/main/model.bin"
# response = requests.get(URL)
# open("model.bin", "wb").write(response.content)
DATASET_LOCATION = "" #
MODEL_PATH = "model.bin"
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
# 7 EPOCH Version
BERT_PATH = "FFZG-cleopatra/bert-emoji-latvian-twitter"
# TODO check if lower casing is required
# BertTokenizer
TOKENIZER = transformers.BertTokenizer.from_pretrained(
BERT_PATH,
do_lower_case=True
)
####################################################################################################################################