L1- OpenCompass 评测 InternLM-1.8B 实践

一、环境配置

1.创建开发机和 conda 环境

   创建开发机界面选择镜像为 Cuda11.7-conda，并选择 GPU 为10% A100

2.安装面向GPU的环境

conda create -n opencompass python=3.10
conda activate opencompass
conda install pytorch==2.1.2 torchvision==0.16.2 torchaudio==2.1.2 pytorch-cuda=12.1 -c pytorch -c nvidia -y

# 注意：一定要先 cd /root
cd /root
git clone -b 0.2.4 https://github.com/open-compass/opencompass
cd opencompass
pip install -e .

apt-get update
apt-get install cmake
pip install -r requirements.txt
pip install protobuf

 

 

 

 

 

 

 

二、数据准备

1.评测数据集

cp /share/temp/datasets/OpenCompassData-core-20231110.zip /root/opencompass/
unzip OpenCompassData-core-20231110.zip 

 

在 OpenCompass 下会看到data文件夹 

 

2.InternLM和ceval 相关的配置文件

python tools/list_configs.py internlm ceval 

列出所有跟 InternLM 及 C-Eval 相关的配置 

 

三、启动评测 (10% A100 8GB 资源)

1.使用命令行配置参数法进行评测

①打开 opencompass文件夹

   configs/models/hf_internlm/的hf_internlm2_chat_1_8b.py ,贴入代码

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),
    )
]

 

②以 --debug 模式启动评估 ，检查是否存在问题

#环境变量配置
export MKL_SERVICE_FORCE_INTEL=1
#或
export MKL_THREADING_LAYER=GNU

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

 

 

③命令解析

python run.py
--datasets ceval_gen \ # 数据集准备
--models hf_internlm2_chat_1_8b \  # 模型准备
--debug 

 

④一切正常，屏幕显示 

[2024-08-09 16:48:07,016] [opencompass.openicl.icl_inferencer.icl_gen_inferencer] [INFO] Starting inference process...

⑤评测完成

 

2.使用配置文件修改参数法进行评测

①在configs文件夹下创建eval_tutorial_demo.py

cd /root/opencompass/configs
touch eval_tutorial_demo.py     

 

②打开eval_tutorial_demo.py 贴入代码 

from mmengine.config import read_base

with read_base():
    from .datasets.ceval.ceval_gen import ceval_datasets
    from .models.hf_internlm.hf_internlm2_chat_1_8b import models as hf_internlm2_chat_1_8b_models

datasets = ceval_datasets
models = hf_internlm2_chat_1_8b_models

 

③将配置文件的路径传递给 run.py

cd /root/opencompass
python run.py configs/eval_tutorial_demo.py --debug 

④一切正常，屏幕显示 

⑤评测完成