-
{monologg/koelectra-small-v3-finetuned-korquad
"epoch": 7.0,
"eval_samples": 484,
"exact_match": 53.75,
"f1": 62.69266917293233
}
{monologg/koelectra-small-v2-distilled-korquad…
-
Like the previous task, can I download korquad-open-ldbd3 dataset?
-
Traceback (most recent call last):
File "run_squad.py", line 78, in
(),
File "run_squad.py", line 77, in
(tuple(conf.pretrained_config_archive_map.keys()) for conf in (BertConfig, Ro…
-
```
python3 run_squad.py --model_type kobert \
--model_name_or_path monologg/kobert \
--output_dir models \
--data_dir data \
…
-
### 현재 상황과 문제점
본 Task에서 데이터 증강은 Question과 Answer에 대한 증강이 필요하다.
Context와 Answer를 고정시키고 Question을 생성하는 Question Generation 방법으로 Question과 Answer Pair를 생성할 수 있다.
### 개선 제안 사항
1. 모델 학습을 통한 Question 생성…
-
### TODO
- [x] KorQuAD로 fine-tuning하기
- [x] #1 에서 제공되는 test 모듈로 테스트하기
-
```
python train.py --base_config korquad/bidaf
```
```
2021-04-24 16:41:17,090 (trainer.py:356): [INFO] - # Train Mode.
2021-04-24 16:41:17,519 (trainer.py:389): [INFO] - Start - Batch Loss:…
-
# 목표
다음을 수행해 성능을 비교하는 것이 실험의 목표입니다.
- [x] KorQuAD-Question-Generation를 사용한 Augmentation
- [x] Pororo를 사용한 Translation
# 결과
### 1차 시도
||Basic|Gen|Pororo|
|:---:|:---:|:---:|:---:|
|LB|34.…
-
파인튜닝.py를 사용했는데 checkpoint 디렉터리에 저장이 되었습니다.
파일에는 adapter_model.bin , optimizer.pt , rng_state.pth , scheduler.pt , training_args.bin 이 있는데
학습 완료된 모델 가중치 파일이 adapter_model.bin이 맞나요?
-