Error when merging Lora weights

[hoangdzung]

Thank you so much for your work. I'm trying to run sh ./dbgpt_hub/scripts/export_merge.sh, but getting the following error. Can you upload the latest script? Thanks.

Traceback (most recent call last):
  File "dbgpt_hub/train/export_model.py", line 14, in <module>
    main()
  File "dbgpt_hub/train/export_model.py", line 10, in main
    export_model()
  File "/home/DB-GPT-Hub/dbgpt_hub/llm_base/model_trainer.py", line 404, in export_model
    model_args, _, training_args, finetuning_args, _ = get_train_args(args, data_args_init=False)
  File "/home/DB-GPT-Hub/dbgpt_hub/llm_base/config_parser.py", line 111, in get_train_args
    ) = parse_train_args(args)
  File "/home/DB-GPT-Hub/dbgpt_hub/llm_base/config_parser.py", line 83, in parse_train_args
    return _parse_args(parser, args)
  File "/home/DB-GPT-Hub/dbgpt_hub/llm_base/config_parser.py", line 62, in _parse_args
    return parser.parse_args_into_dataclasses()
  File "/home/anaconda3/envs/dbgpt/lib/python3.8/site-packages/transformers/hf_argparser.py", line 347, in parse_args_into_dataclasses
    raise ValueError(f"Some specified arguments are not used by the HfArgumentParser: {remaining_args}")
ValueError: Some specified arguments are not used by the HfArgumentParser: ['-checkpoint_dir', '/home/CodeLlama-13b-sql-lora/']

[junewgl]

Can you provide specific information about your export_merge.sh? Are the weights based on our train_sft.sh? You can directly use the huggingface weights we uploaded. @hoangdzung