선택지(보기)를 늘렸을 때 오류 발생

[hsmin9809]

Traceback (most recent call last):

  File "/data/ephemeral/home/camper/seongmin/level2-nlp-generationfornlp-nlp-02-lv3/code/main.py", line 87, in <module>
    main()
    └ <function main at 0x7f120f03b010>

> File "/data/ephemeral/home/camper/seongmin/level2-nlp-generationfornlp-nlp-02-lv3/code/main.py", line 58, in main
    trained_model = trainer.train()
                    │       └ <function CustomTrainer.train at 0x7f11105fa9e0>
                    └ <trainer.CustomTrainer object at 0x7f11ef4ea8f0>

  File "/data/ephemeral/home/camper/seongmin/level2-nlp-generationfornlp-nlp-02-lv3/code/trainer.py", line 21, in train
    trainer.train()
    │       └ <function Trainer.train at 0x7f11114b69e0>
    └ <trl.trainer.sft_trainer.SFTTrainer object at 0x7f110d1823e0>

  File "/data/ephemeral/home/camper/.local/lib/python3.10/site-packages/transformers/trainer.py", line 2123, in train
    return inner_training_loop(
           └ functools.partial(<bound method Trainer._inner_training_loop of <trl.trainer.sft_trainer.SFTTrainer object at 0x7f110d1823e0>...
  File "/data/ephemeral/home/camper/.local/lib/python3.10/site-packages/transformers/trainer.py", line 2548, in _inner_training_loop
    self._maybe_log_save_evaluate(tr_loss, grad_norm, model, trial, epoch, ignore_keys_for_eval)
    │    │                        │        │          │      │      │      └ None
    │    │                        │        │          │      │      └ 1
    │    │                        │        │          │      └ None
    │    │                        │        │          └ PeftModelForCausalLM(
    │    │                        │        │              (base_model): LoraModel(
    │    │                        │        │                (model): ExaoneForCausalLM(
    │    │                        │        │                  (transformer): ExaoneModel(
    │    │                        │        │                    (w...
    │    │                        │        └ tensor(0.0019, device='cuda:0')
    │    │                        └ tensor(0., device='cuda:0')
    │    └ <function Trainer._maybe_log_save_evaluate at 0x7f11114b6dd0>
    └ <trl.trainer.sft_trainer.SFTTrainer object at 0x7f110d1823e0>
  File "/data/ephemeral/home/camper/.local/lib/python3.10/site-packages/transformers/trainer.py", line 3004, in _maybe_log_save_evaluate
    metrics = self._evaluate(trial, ignore_keys_for_eval)
              │    │         │      └ None
              │    │         └ None
              │    └ <function Trainer._evaluate at 0x7f11114b6d40>
              └ <trl.trainer.sft_trainer.SFTTrainer object at 0x7f110d1823e0>
  File "/data/ephemeral/home/camper/.local/lib/python3.10/site-packages/transformers/trainer.py", line 2958, in _evaluate
    metrics = self.evaluate(ignore_keys=ignore_keys_for_eval)
              │    │                    └ None
              │    └ <function Trainer.evaluate at 0x7f11114b7ac0>
              └ <trl.trainer.sft_trainer.SFTTrainer object at 0x7f110d1823e0>
  File "/data/ephemeral/home/camper/.local/lib/python3.10/site-packages/transformers/trainer.py", line 3975, in evaluate
    output = eval_loop(
             └ <bound method Trainer.evaluation_loop of <trl.trainer.sft_trainer.SFTTrainer object at 0x7f110d1823e0>>
  File "/data/ephemeral/home/camper/.local/lib/python3.10/site-packages/transformers/trainer.py", line 4264, in evaluation_loop
    metrics = self.compute_metrics(
              │    └ <bound method CustomTrainer._compute_metrics of <trainer.CustomTrainer object at 0x7f11ef4ea8f0>>
              └ <trl.trainer.sft_trainer.SFTTrainer object at 0x7f110d1823e0>

  File "/data/ephemeral/home/camper/seongmin/level2-nlp-generationfornlp-nlp-02-lv3/code/trainer.py", line 107, in _compute_metrics
    labels = [self.int_output_map[x] for x in labels]
              │    │                          └ ['1', '2', '2', '3', '4', '3', '6', '2', '1', '6', '4', '3', '5', '2', '6', '5', '4', '3', '2', '1', '4', '1', '2', '4', '2',...
              │    └ {'1': 0, '2': 1, '3': 2, '4': 3, '5': 4}
              └ <trainer.CustomTrainer object at 0x7f11ef4ea8f0>

  File "/data/ephemeral/home/camper/seongmin/level2-nlp-generationfornlp-nlp-02-lv3/code/trainer.py", line 107, in <listcomp>
    labels = [self.int_output_map[x] for x in labels]
              │    │              │      └ '6'
              │    │              └ '6'
              │    └ {'1': 0, '2': 1, '3': 2, '4': 3, '5': 4}
              └ <trainer.CustomTrainer object at 0x7f11ef4ea8f0>

KeyError: '6'

1 epoch가 끝나고 eval이 완료된 직후에 이런 오류가 발생하는데, 발생 원인을 모르겠음...

혹시 문제의 답이 6개인 경우 보기가 6개가 아닌 에러를 예상하여 체크하였으나, 모두 보기가 6개인 것을 확인하였다.

[Sujinkim-625]

self.int_output_map = {"1": 0, "2": 1, "3": 2, "4": 3, "5": 4}

를 확인해봐요!

[hsmin9809]

self.int_output_map = {"1": 0, "2": 1, "3": 2, "4": 3, "5": 4, "6": 5}

로 수정하니 해결 완료! 감사감사~