fix: 'DataArguments' object has no attribute 'dataset_config_name'

[KINNNNNNG]

When I was training with chat/train.py, I reported the following error after training:

Traceback (most recent call last):
  File "train.py", line 345, in <module>
    main()
  File "train.py", line 313, in main
    if data_args.dataset_config_name is not None:
AttributeError: 'DataArguments' object has no attribute 'dataset_config_name'
WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 2461 closing signal SIGTERM
WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 2462 closing signal SIGTERM
ERROR:torch.distributed.elastic.multiprocessing.api:failed (exitcode: 1) local_rank: 0 (pid: 2460) of binary: /usr/bin/python
Traceback (most recent call last):
  File "/usr/local/bin/torchrun", line 8, in <module>
    sys.exit(main())
  File "/usr/local/lib/python3.7/dist-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
    return f(*args, **kwargs)
  File "/usr/local/lib/python3.7/dist-packages/torch/distributed/run.py", line 762, in main
    run(args)
  File "/usr/local/lib/python3.7/dist-packages/torch/distributed/run.py", line 756, in run
    )(*cmd_args)
  File "/usr/local/lib/python3.7/dist-packages/torch/distributed/launcher/api.py", line 132, in __call__
    return launch_agent(self._config, self._entrypoint, list(args))
  File "/usr/local/lib/python3.7/dist-packages/torch/distributed/launcher/api.py", line 248, in launch_agent
    failures=result.failures,
torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
============================================================
train.py FAILED
------------------------------------------------------------
Failures:
  <NO_OTHER_FAILURES>
------------------------------------------------------------
Root Cause (first observed failure):
[0]:
  time      : 2023-06-06_07:00:57
  host      : ee11o7smse510-0
  rank      : 0 (local_rank: 0)
  exitcode  : 1 (pid: 2460)
  error_file: <N/A>
  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html

I looked at the config.py file and found that DataArguments was missing the dataset_config_name statement, I wrote a test.py for testing, After I added dataset_config_name, the test passed.

import logging
import math
import os
import random
import sys
from itertools import chain

import datasets
import torch
import transformers
from config import DataArguments, ModelArguments, TrainingArguments
from datasets import load_dataset
from dialogues import get_dialogue_template, mask_user_labels, prepare_dialogue
from transformers import (AutoModelForCausalLM, AutoTokenizer, Trainer,
                          default_data_collator, set_seed)
from transformers.testing_utils import CaptureLogger
from transformers.trainer_utils import get_last_checkpoint
from utils import StarChatArgumentParser, hf_login

if __name__ == '__main__':

    parser = StarChatArgumentParser((ModelArguments, DataArguments, TrainingArguments))
    if len(sys.argv) == 2 and sys.argv[1].endswith(".yaml"):
        # If we pass only one argument to the script and it's the path to a YAML file,
        # let's parse it to get our arguments.
        model_args, data_args, training_args = parser.parse_yaml_file(os.path.abspath(sys.argv[1]))
    # parse command line args and yaml file
    elif len(sys.argv) > 2 and sys.argv[1].endswith(".yaml"):
        model_args, data_args, training_args = parser.parse_yaml_and_args(os.path.abspath(sys.argv[1]), sys.argv[2:])
    # parse command line args only
    else:
        model_args, data_args, training_args = parser.parse_args_into_dataclasses()

    kwargs = {}
    if data_args.dataset_name is not None:
        kwargs["dataset_tags"] = data_args.dataset_name
        if data_args.dataset_config_name is not None:
            kwargs["dataset_args"] = data_args.dataset_config_name
            kwargs["dataset"] = f"{data_args.dataset_name} {data_args.dataset_config_name}"
        else:
            kwargs["dataset"] = data_args.dataset_name
            kwargs["dataset_args"] = "default"


    print(data_args.dataset_config_name is not None)
    print(kwargs)

[davide221]

Got the same problem and fixed in the same way

[lvwerra]

Looks good to me, I'll let @ArmelRandy do a final check.