def __init__(self, hparams):
super(T5FineTuner, self).__init__()
self.hparams = hparams
self.model = MT5ForConditionalGeneration.from_pretrained(
hparams.model_name_or_path)
self.tokenizer = MT5TokenizerFast.from_pretrained(
hparams.tokenizer_name_or_path)
self.model.get_output_embeddings().weight.requires_grad = False
self.model.get_input_embeddings().weight.requires_grad = False
import torch
from torch.utils.data import DataLoader
from tqdm.auto import tqdm
from sklearn import metrics
import pytorch_lightning as pl
from transformers import MT5TokenizerFast,MT5ForConditionalGeneration
from trainudp import PostagDataset,T5FineTuner,args
tokenizer = MT5TokenizerFast.from_pretrained('mt5tokenizer')
dataset = PostagDataset(tokenizer, 'udp', 'test', max_len=200)
loader = DataLoader(dataset, batch_size=8, num_workers=1)
model = T5FineTuner(args)
model.model=model.model.from_pretrained('udp/result')
model.model.eval()
outputs = []
targets = []
f1_list=[]
if __name__ == '__main__':
torch.multiprocessing.freeze_support()
device = "cuda"
model = model.to(device)
truep, ref, pred = 0, 0, 0
for batch in tqdm(loader):
outs = model.model.generate(input_ids=batch['source_ids'].cuda(),
attention_mask=batch['source_mask'].cuda(),
max_length=140)
'''pred = [tokenizer.decode(ids) for ids in outs]