L1-OpenCompass 评测 InternLM-1.8B 实践

一、数据准备

解压后在opencompass文件夹下面看到数据集

可以查看到所有跟 InternLM 及 C-Eval 相关的配置

二、启动评测

找到hf_internlm2_chat_1_8b.py

from opencompass.models import HuggingFaceCausalLM


models = [
    dict(
        type=HuggingFaceCausalLM,
        abbr='internlm2-1.8b-hf',
        path="/share/new_models/Shanghai_AI_Laboratory/internlm2-chat-1_8b",
        tokenizer_path='/share/new_models/Shanghai_AI_Laboratory/internlm2-chat-1_8b',
        model_kwargs=dict(
            trust_remote_code=True,
            device_map='auto',
        ),
        tokenizer_kwargs=dict(
            padding_side='left',
            truncation_side='left',
            use_fast=False,
            trust_remote_code=True,
        ),
        max_out_len=100,
        min_out_len=1,
        max_seq_len=2048,
        batch_size=8,
        run_cfg=dict(num_gpus=1, num_procs=1),
    )
]

配置环境变量

export MKL_THREADING_LAYER=GNU

运行程序

python run.py --datasets ceval_gen --models hf_internlm2_chat_1_8b --debug

得到如图所示的评测结果，用时约3h