使用InstructionWild数据中的seed_prompts_en.jsonl数据集对llama-7b模型进行预训练报错

[13416157913]

执行命令是：torchrun --standalone --nproc_per_node=1 train_sft.py --pretrain "/home/llm-deploy/ColossalAI/llama-7b" --model 'llama' --strategy colossalai_zero2 --log_interval 10 --save_path /home/llm-deploy/ColossalAI/output/llama-7B --dataset /home/llm-deploy/ColossalAI/InstructionWild/data/seed_prompts_en.jsonl --batch_size 1 --lr 2e-5 --max_epochs 1

报错信息如下：

[CWHer]

It seems that you used the wrong dataset. 🤔

[CWHer]

The dataset should follow the following format,

[
    {
        "instruction": "Give three tips for staying healthy.",
        "input": "",
        "output": "1.Eat a balanced diet and make sure to include plenty of fruits and vegetables. \n2. Exercise regularly to keep your body active and strong. \n3. Get enough sleep and maintain a consistent sleep schedule."
    },
...

[flybird11111]

执行命令是：torchrun --standalone --nproc_per_node=1 train_sft.py --pretrain "/home/llm-deploy/ColossalAI/llama-7b" --model 'llama' --strategy colossalai_zero2 --log_interval 10 --save_path /home/llm-deploy/ColossalAI/output/llama-7B --dataset /home/llm-deploy/ColossalAI/InstructionWild/data/seed_prompts_en.jsonl --batch_size 1 --lr 2e-5 --max_epochs 1

报错信息如下：

The training data is contained in these two files.