Skip to content

询问下代码 #1

@tzbs

Description

@tzbs

大神,您好,下面代码好像有问题呀!
from utils.merge import merge_llm_with_lora
这个找不到在哪?

parser.add_argument("--dataset_name", type=str, default="./data/alpaca_gpt4_data.json")
def create_datasets(tokenizer, args):
train_json_path = os.path.join(args.dataset_name, "train/empathetic_dialogue_train.json")
train_data = load_dataset("json", data_files=train_json_path, split="train")
train_data = train_data.shuffle(seed=args.seed)

val_json_path = os.path.join(args.dataset_name, "empathetic_dialogue_valid.json")
valid_data = load_dataset("json", data_files=val_json_path, split="train")

supervised_finetuning_cot.py不是用ESD-CoT数据集对预训练模型进行微调吗,在推理的时候,输入对话上下文,会输出五元组。这里的训练集和验证集是empathetic_dialogue_train.json、empathetic_dialoge.valid.json吗?valid_data = load_dataset("json", data_files=val_json_path, split="train"),这里划分是不是不对啊?

恳求大神指教啊!

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions