ascend-npu，执行模型评估脚本时，出现IndexError: list index out of range

[doit-5618]

Reminder

• I have read the above rules and searched the existing issues.

System Info

(llamafactory2) root@unicom:/mnt/llm-technology/LLaMA-Factory# llamafactory-cli env
[2025-04-02 17:23:39,239] [INFO] [real_accelerator.py:222:get_accelerator] Setting ds_accelerator to npu (auto detect)

• llamafactory version: 0.9.3.dev0
• Platform: Linux-5.15.0-131-generic-aarch64-with-glibc2.35
• Python version: 3.10.16
• PyTorch version: 2.4.0 (NPU)
• Transformers version: 4.49.0
• Datasets version: 3.3.2
• Accelerate version: 1.4.0
• PEFT version: 0.12.0
• TRL version: 0.9.6
• NPU type: Ascend910B2
• CANN version: 8.0.0
• DeepSpeed version: 0.16.4
• Git commit: 555b71a

Reproduction

我参考https://ascend.github.io/docs/sources/llamafactory/example.html#benchmark，使用下面的脚本进行模型评估。
llamafactory-cli eval \
    --model_name_or_path /mnt/llm-technology/LLaMA-Factory/saves/sft-41-full \
    --template fewshot \
    --task mmlu \
    --split validation \
    --lang en \
    --n_shot 5 \
    --batch_size 1
报错：llamafactory-cli: error: ambiguous option: --split could match --split_special_tokens, --split-special-tokens
随后我删除--split，报错IndexError: list index out of range 。
可以执行训练、推理。

Others

No response

[Feihustc]

me too

[doit-5618]

me too

改为：--task ceval_validation