This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
""" | |
Тренировка модели болталки Axioma на FRED T5 для проекта https://github.com/Koziev/chatbot | |
Эксперимент с файнтюном: токены истории диалога не включаем в backprop, присваивая соответствующим целям (labels) значение -100 | |
Прочие хинты по тренировке: https://kelijah.livejournal.com/315826.html | |
""" | |
import os | |
import json | |
import sys | |
import io |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
import os | |
import argparse | |
import torch | |
import transformers | |
from transformers import T5Config | |
if __name__ == '__main__': | |
proj_dir = os.path.expanduser('~/polygon/chatbot') |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
""" | |
Определение склонности моделей rugpt к запоминанию обучающего датасета | |
""" | |
import collections | |
import os | |
import json | |
import itertools | |
import sys | |
import argparse | |
import random |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
""" | |
Определение склонности моделей rugpt к запоминанию обучающего датасета | |
""" | |
import collections | |
import os | |
import json | |
import itertools | |
import sys | |
import argparse | |
import random |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
import random | |
import os | |
import torch | |
import transformers | |
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer | |
import deepspeed | |
import matplotlib.pyplot as plt | |
OlderNewer