You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
150 lines
4.9 KiB
150 lines
4.9 KiB
import os
|
|
import sys
|
|
|
|
os.environ["HF_ENDPOINT"] = "https://hf-mirror.com"
|
|
sys.path.append('/root/hjt/md_bayesian_er_ditto/')
|
|
import pdb
|
|
import pickle
|
|
import torch
|
|
import json
|
|
import numpy as np
|
|
import random
|
|
from setting import *
|
|
from colorama import Fore
|
|
from argparse import Namespace
|
|
import ConfigSpace
|
|
from ConfigSpace import Configuration
|
|
from ditto.matcher import set_seed, predict, tune_threshold, load_model
|
|
from ConfigSpace.read_and_write import json as csj
|
|
from ditto.ditto_light.dataset import DittoDataset
|
|
from ditto.ditto_light.summarize import Summarizer
|
|
from ditto.ditto_light.knowledge import *
|
|
from ditto.ditto_light.ditto import train
|
|
|
|
|
|
def matching(hpo_config):
|
|
print(Fore.BLUE + f'Config: {hpo_config}')
|
|
|
|
with open(md_output_dir + "/mds.pickle", "rb") as file:
|
|
md_list = pickle.load(file)
|
|
|
|
hp = Namespace()
|
|
hp.task = directory_path.replace('/root/hjt/DeepMatcher Dataset/', '')
|
|
|
|
# only a single task for baseline
|
|
task = hp.task
|
|
# load task configuration
|
|
configs = json.load(open('../ditto/configs.json'))
|
|
configs = {conf['name']: conf for conf in configs}
|
|
config = configs[task]
|
|
|
|
config['trainset'] = '/root/hjt/md_bayesian_er_ditto/ditto/' + config['trainset']
|
|
config['validset'] = '/root/hjt/md_bayesian_er_ditto/ditto/' + config['validset']
|
|
config['testset'] = '/root/hjt/md_bayesian_er_ditto/ditto/' + config['testset']
|
|
|
|
trainset = config['trainset']
|
|
validset = config['validset']
|
|
testset = config['testset']
|
|
|
|
hp.run_id = 0
|
|
hp.batch_size = hpo_config['batch_size']
|
|
hp.max_len = hpo_config['max_len']
|
|
hp.lr = 3e-5
|
|
hp.n_epochs = 20
|
|
# hp.finetuning
|
|
hp.save_model = True
|
|
hp.input_path = config['testset']
|
|
hp.output_path = '/root/autodl-tmp/output/matched_small.jsonl'
|
|
|
|
hp.logdir = '/root/autodl-tmp/checkpoints/'
|
|
hp.checkpoint_path = '/root/autodl-tmp/checkpoints/'
|
|
hp.lm = hpo_config['language_model']
|
|
hp.fp16 = hpo_config['half_precision_float']
|
|
hp.da = hpo_config['data_augmentation']
|
|
hp.alpha_aug = 0.8
|
|
hp.dk = None
|
|
hp.summarize = hpo_config['summarize']
|
|
hp.size = None
|
|
hp.use_gpu = True
|
|
|
|
seed = hp.run_id
|
|
random.seed(seed)
|
|
np.random.seed(seed)
|
|
torch.manual_seed(seed)
|
|
if torch.cuda.is_available():
|
|
torch.cuda.manual_seed_all(seed)
|
|
|
|
# create the tag of the run
|
|
run_tag = '%s_lm=%s_da=%s_dk=%s_su=%s_size=%s_id=%d' % (task, hp.lm, hp.da,
|
|
hp.dk, hp.summarize, str(hp.size), hp.run_id)
|
|
run_tag = run_tag.replace('/', '_')
|
|
|
|
# summarize the sequences up to the max sequence length
|
|
if hp.summarize:
|
|
summarizer = Summarizer(config, lm=hp.lm)
|
|
trainset = summarizer.transform_file(trainset, max_len=hp.max_len)
|
|
validset = summarizer.transform_file(validset, max_len=hp.max_len)
|
|
testset = summarizer.transform_file(testset, max_len=hp.max_len)
|
|
|
|
# load train/dev/test sets
|
|
train_dataset = DittoDataset(trainset,
|
|
lm=hp.lm,
|
|
max_len=hp.max_len,
|
|
size=hp.size,
|
|
da=hp.da)
|
|
valid_dataset = DittoDataset(validset, lm=hp.lm)
|
|
test_dataset = DittoDataset(testset, lm=hp.lm)
|
|
|
|
# train and evaluate the model
|
|
train(train_dataset,
|
|
valid_dataset,
|
|
test_dataset,
|
|
run_tag, hp)
|
|
|
|
set_seed(123)
|
|
config, model = load_model(hp.task, hp.checkpoint_path,
|
|
hp.lm, hp.use_gpu, hp.fp16)
|
|
|
|
summarizer = dk_injector = None
|
|
pdb.set_trace()
|
|
if hp.summarize:
|
|
summarizer = Summarizer(config, hp.lm)
|
|
|
|
# tune threshold
|
|
threshold = tune_threshold(config, model, hp)
|
|
|
|
# run prediction
|
|
predict(hp.input_path, hp.output_path, config, model,
|
|
summarizer=summarizer,
|
|
max_len=hp.max_len,
|
|
lm=hp.lm,
|
|
dk_injector=dk_injector,
|
|
threshold=threshold)
|
|
# todo indicators
|
|
# write results
|
|
# interpretability
|
|
indicators = {}
|
|
return indicators
|
|
|
|
|
|
# todo ml_er function
|
|
def ml_er(config: Configuration):
|
|
indicators = matching(config)
|
|
output_path = er_output_dir + "/eval_result.txt"
|
|
with open(output_path, 'w') as _f:
|
|
_f.write('F1:' + str(indicators["F1"]) + '\n')
|
|
_f.write('interpretability:' + str(indicators['interpretability']) + '\n')
|
|
_f.write('performance:' + str(indicators['performance']) + '\n')
|
|
|
|
|
|
if __name__ == '__main__':
|
|
if os.path.isfile(hpo_output_dir + "/incumbent.json"):
|
|
with open(hpo_output_dir + "/configspace.json", 'r') as f:
|
|
dict_configspace = json.load(f)
|
|
str_configspace = json.dumps(dict_configspace)
|
|
configspace = csj.read(str_configspace)
|
|
with open(hpo_output_dir + "/incumbent.json", 'r') as f:
|
|
dic = json.load(f)
|
|
configuration = ConfigSpace.Configuration(configspace, values=dic)
|
|
ml_er(configuration)
|