本文详细介绍了OpenCompass，一个用于评测主流大模型的开源平台，涵盖了模型选型、能力测试（如知识、推理、对话等）以及实际操作步骤，包括数据准备、配置和命令行评测。OpenCompass支持HuggingFace模型和多种评测指标，为模型开发者和评估者提供了全面的评估工具。
视频地址：
OpenCompass 大模型评测_哔哩哔哩_bilibili
MarkDown：
https://github.com/InternLM/tutorial/blob/main/opencompass/opencompass_tutorial.md
Repo:
https://github.com/open-compass/opencompass
1.为什么要评测
- 模型选型
- 模型能力提升
- 真实应用场景效果评测
2.要测什么内容
- 知识、推理、语言
- 长文本、智能体、多轮对话
- 情感、认知、价值观
3.怎么评测大语言模型
-
自动化客观评测
问答、多选、判断、分类等
-
人机交互评测
主观评价、模型评价
-
基于大模型的大模型评测
基座模型、对话模型等
-
提示词工程
4.OpenCompass
——主流大模型评测框架
OpenCompass开源评测平台架构
OpenCompass评测流水线设计
大模型能力对比
5.OpenCompass评测实战
5.1 基础环境
conda create --name opencompass --clone=/root/share/conda_envs/internlm-base
conda activate opencompass
git clone https://github.com/open-compass/opencompass
cd opencompass
pip install -e .
有部分第三方功能,如代码能力基准测试 Humaneval 以及 Llama格式的模型评测,可能需要额外步骤才能正常运行
5.2 数据准备
# 解压评测数据集到 data/ 处
cp /share/temp/datasets/OpenCompassData-core-20231110.zip /root/opencompass/
unzip OpenCompassData-core-20231110.zip
# 将会在opencompass下看到data文件夹
5.3 查看支持的数据集和模型
# 列出所有跟 internlm 及 ceval 相关的配置
python tools/list_configs.py internlm ceval
将会看到
+--------------------------+--------------------------------------------------------+
| Model | Config Path |
|--------------------------+--------------------------------------------------------|
| hf_internlm_20b | configs/models/hf_internlm/hf_internlm_20b.py |
| hf_internlm_7b | configs/models/hf_internlm/hf_internlm_7b.py |
| hf_internlm_chat_20b | configs/models/hf_internlm/hf_internlm_chat_20b.py |
| hf_internlm_chat_7b | configs/models/hf_internlm/hf_internlm_chat_7b.py |
| hf_internlm_chat_7b_8k | configs/models/hf_internlm/hf_internlm_chat_7b_8k.py |
| hf_internlm_chat_7b_v1_1 | configs/models/hf_internlm/hf_internlm_chat_7b_v1_1.py |
| internlm_7b | configs/models/internlm/internlm_7b.py |
| ms_internlm_chat_7b_8k | configs/models/ms_internlm/ms_internlm_chat_7b_8k.py |
+--------------------------+--------------------------------------------------------+
+----------------------------+------------------------------------------------------+
| Dataset | Config Path |
|----------------------------+------------------------------------------------------|
| ceval_clean_ppl | configs/datasets/ceval/ceval_clean_ppl.py |
| ceval_gen | configs/datasets/ceval/ceval_gen.py |
| ceval_gen_2daf24 | configs/datasets/ceval/ceval_gen_2daf24.py |
| ceval_gen_5f30c7 | configs/datasets/ceval/ceval_gen_5f30c7.py |
| ceval_ppl | configs/datasets/ceval/ceval_ppl.py |
| ceval_ppl_578f8d | configs/datasets/ceval/ceval_ppl_578f8d.py |
| ceval_ppl_93e5ce | configs/datasets/ceval/ceval_ppl_93e5ce.py |
| ceval_zero_shot_gen_bd40ef | configs/datasets/ceval/ceval_zero_shot_gen_bd40ef.py |
+----------------------------+------------------------------------------------------+
5.4 启动评测
确保按照上述步骤正确安装 OpenCompass 并准备好数据集后，可以通过以下命令评测 InternLM-Chat-7B 模型在 C-Eval 数据集上的性能。由于 OpenCompass 默认并行启动评估过程，我们可以在第一次运行时以 --debug 模式启动评估，并检查是否存在问题。在 --debug 模式下，任务将按顺序执行，并实时打印输出。
python run.py --datasets ceval_gen --hf-path /share/temp/model_repos/internlm-chat-7b/ --tokenizer-path /share/temp/model_repos/internlm-chat-7b/ --tokenizer-kwargs padding_side='left' truncation='left' trust_remote_code=True --model-kwargs trust_remote_code=True device_map='auto' --max-seq-len 2048 --max-out-len 16 --batch-size 4 --num-gpus 1 --debug
命令解析
--datasets ceval_gen \
--hf-path /share/temp/model_repos/internlm-chat-7b/ \ # HuggingFace 模型路径
--tokenizer-path /share/temp/model_repos/internlm-chat-7b/ \ # HuggingFace tokenizer 路径（如果与模型路径相同，可以省略）
--tokenizer-kwargs padding_side='left' truncation='left' trust_remote_code=True \ # 构建 tokenizer 的参数
--model-kwargs device_map='auto' trust_remote_code=True \ # 构建模型的参数
--max-seq-len 2048 \ # 模型可以接受的最大序列长度
--max-out-len 16 \ # 生成的最大 token 数
--batch-size 2 \ # 批量大小
--num-gpus 1 # 运行模型所需的 GPU 数量
--debug
如果一切正常，您应该看到屏幕上显示 “Starting inference process”：
[2024-01-12 18:23:55,076] [opencompass.openicl.icl_inferencer.icl_gen_inferencer] [INFO] Starting inference process...
评测完成后，将会看到：
dataset version metric mode opencompass.models.huggingface.HuggingFace_model_repos_internlm-chat-7b
---------------------------------------------- --------- ------------- ------ -------------------------------------------------------------------------
ceval-computer_network db9ce2 accuracy gen 31.58
ceval-operating_system 1c2571 accuracy gen 36.84
ceval-computer_architecture a74dad accuracy gen 28.57
ceval-college_programming 4ca32a accuracy gen 32.43
ceval-college_physics 963fa8 accuracy gen 26.32
ceval-college_chemistry e78857 accuracy gen 16.67
ceval-advanced_mathematics ce03e2 accuracy gen 21.05
ceval-probability_and_statistics 65e812 accuracy gen 38.89
ceval-discrete_mathematics e894ae accuracy gen 18.75
ceval-electrical_engineer ae42b9 accuracy gen 35.14
ceval-metrology_engineer ee34ea accuracy gen 50
ceval-high_school_mathematics 1dc5bf accuracy gen 22.22
ceval-high_school_physics adf25f accuracy gen 31.58
ceval-high_school_chemistry 2ed27f accuracy gen 15.79
ceval-high_school_biology 8e2b9a accuracy gen 36.84
ceval-middle_school_mathematics bee8d5 accuracy gen 26.32
ceval-middle_school_biology 86817c accuracy gen 61.9
ceval-middle_school_physics 8accf6 accuracy gen 63.16
ceval-middle_school_chemistry 167a15 accuracy gen 60
ceval-veterinary_medicine b4e08d accuracy gen 47.83
ceval-college_economics f3f4e6 accuracy gen 41.82
ceval-business_administration c1614e accuracy gen 33.33
ceval-marxism cf874c accuracy gen 68.42
ceval-mao_zedong_thought 51c7a4 accuracy gen 70.83
ceval-education_science 591fee accuracy gen 58.62
ceval-teacher_qualification 4e4ced accuracy gen 70.45
ceval-high_school_politics 5c0de2 accuracy gen 26.32
ceval-high_school_geography 865461 accuracy gen 47.37
ceval-middle_school_politics 5be3e7 accuracy gen 52.38
ceval-middle_school_geography 8a63be accuracy gen 58.33
ceval-modern_chinese_history fc01af accuracy gen 73.91
ceval-ideological_and_moral_cultivation a2aa4a accuracy gen 63.16
ceval-logic f5b022 accuracy gen 31.82
ceval-law a110a1 accuracy gen 25
ceval-chinese_language_and_literature 0f8b68 accuracy gen 30.43
ceval-art_studies 2a1300 accuracy gen 60.61
ceval-professional_tour_guide 4e673e accuracy gen 62.07
ceval-legal_professional ce8787 accuracy gen 39.13
ceval-high_school_chinese 315705 accuracy gen 63.16
ceval-high_school_history 7eb30a accuracy gen 70
ceval-middle_school_history 48ab4a accuracy gen 59.09
ceval-civil_servant 87d061 accuracy gen 53.19
ceval-sports_science 70f27b accuracy gen 52.63
ceval-plant_protection 8941f9 accuracy gen 59.09
ceval-basic_medicine c409d6 accuracy gen 47.37
ceval-clinical_medicine 49e82d accuracy gen 40.91
ceval-urban_and_rural_planner 95b885 accuracy gen 45.65
ceval-accountant 002837 accuracy gen 26.53
ceval-fire_engineer bc23f5 accuracy gen 22.58
ceval-environmental_impact_assessment_engineer c64e2d accuracy gen 64.52
ceval-tax_accountant 3a5e3c accuracy gen 34.69
ceval-physician 6e277d accuracy gen 40.82
ceval-stem - naive_average gen 35.09
ceval-social-science - naive_average gen 52.79
ceval-humanities - naive_average gen 52.58
ceval-other - naive_average gen 44.36
ceval-hard - naive_average gen 23.91
ceval - naive_average gen 44.16
有关 run.py 支持的所有与 HuggingFace 相关的参数，请阅读 评测任务发起
除了通过命令行配置实验外，OpenCompass 还允许用户在配置文件中编写实验的完整配置，并通过 run.py 直接运行它。配置文件是以 Python 格式组织的，并且必须包括 datasets 和 models 字段。
示例测试配置在 configs/eval_demo.py 中。此配置通过 继承机制 引入所需的数据集和模型配置，并以所需格式组合 datasets 和 models 字段。
from mmengine.config import read_base
with read_base():
from .datasets.siqa.siqa_gen import siqa_datasets
from .datasets.winograd.winograd_ppl import winograd_datasets
from .models.opt.hf_opt_125m import opt125m
from .models.opt.hf_opt_350m import opt350m
datasets = [*siqa_datasets, *winograd_datasets]
models = [opt125m, opt350m]
运行任务时，我们只需将配置文件的路径传递给 run.py：
python run.py configs/eval_demo.py
OpenCompass 提供了一系列预定义的模型配置，位于 configs/models 下。以下是与 opt-350m（configs/models/opt/hf_opt_350m.py）相关的配置片段：
# 使用 `HuggingFaceCausalLM` 评估由 HuggingFace 的 `AutoModelForCausalLM` 支持的模型
from opencompass.models import HuggingFaceCausalLM
# OPT-350M
opt350m = dict(
type=HuggingFaceCausalLM,
# `HuggingFaceCausalLM` 的初始化参数
path='facebook/opt-350m',
tokenizer_path='facebook/opt-350m',
tokenizer_kwargs=dict(
padding_side='left',
truncation_side='left',
proxies=None,
trust_remote_code=True),
model_kwargs=dict(device_map='auto'),
# 下面是所有模型的共同参数，不特定于 HuggingFaceCausalLM
abbr='opt350m', # 结果显示的模型缩写
max_seq_len=2048, # 整个序列的最大长度
max_out_len=100, # 生成的最大 token 数
batch_size=64, # 批量大小
run_cfg=dict(num_gpus=1), # 该模型所需的 GPU 数量
)
使用配置时，我们可以通过命令行参数 --models 指定相关文件，或使用继承机制将模型配置导入到配置文件中的 models 列表中。
与模型类似，数据集的配置文件也提供在 configs/datasets 下。用户可以在命令行中使用 --datasets，或通过继承在配置文件中导入相关配置
下面是来自 configs/eval_demo.py 的与数据集相关的配置片段：
from mmengine.config import read_base # 使用 mmengine.read_base() 读取基本配置
with read_base():
# 直接从预设的数据集配置中读取所需的数据集配置
from .datasets.winograd.winograd_ppl import winograd_datasets # 读取 Winograd 配置，基于 PPL（困惑度）进行评估
from .datasets.siqa.siqa_gen import siqa_datasets # 读取 SIQA 配置，基于生成进行评估
datasets = [*siqa_datasets, *winograd_datasets] # 最终的配置需要包含所需的评估数据集列表 'datasets'
数据集配置通常有两种类型：'ppl' 和 'gen'，分别指示使用的评估方法。其中 ppl 表示辨别性评估，gen 表示生成性评估。
此外，configs/datasets/collections 收录了各种数据集集合，方便进行综合评估。OpenCompass 通常使用 base_medium.py 进行全面的模型测试。要复制结果，只需导入该文件，例如：
python run.py --models hf_llama_7b --datasets base_medium
OpenCompass 通常假定运行环境网络是可用的。如果您遇到网络问题或希望在离线环境中运行 OpenCompass，请参阅 FAQ - 网络 - Q1 寻求解决方案。
5.5 可视化评估结果
评估完成后，评估结果表格将打印如下：
dataset version metric mode opt350m opt125m
--------- --------- -------- ------ --------- ---------
siqa e78df3 accuracy gen 21.55 12.44
winograd b6c7ed accuracy ppl 51.23 49.82
所有运行输出将定向到 outputs/demo/ 目录，结构如下：
outputs/default/
├── 20200220_120000
├── 20230220_183030 # 每个实验一个文件夹
│ ├── configs # 用于记录的已转储的配置文件。如果在同一个实验文件夹中重新运行了不同的实验，可能会保留多个配置
│ ├── logs # 推理和评估阶段的日志文件
│ │ ├── eval
│ │ └── infer
│ ├── predictions # 每个任务的推理结果
│ ├── results # 每个任务的评估结果
│ └── summary # 单个实验的汇总评估结果
├── ...
打印评测结果的过程可被进一步定制化，用于输出一些数据集的平均分 (例如 MMLU, C-Eval 等)。
关于评测结果输出的更多介绍可阅读 结果展示。
6.作业
基础作业
命令行评测：
python run.py --datasets ceval_gen --hf-path share/model_repos/internlm2-chat-7b/ --tokenizer-path /share/model_repos/internlm2-chat-7b/ --tokenizer-kwargs padding_side='left' truncation='left' trust_remote_code=True --model-kwargs trust_remote_code=True device_map='auto' --max-seq-len 2048 --max-out-len 16 --batch-size 4 --num-gpus 1 --debug
结果如下：
20240129_013728
tabulate format
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
dataset version metric mode opencompass.models.huggingface.HuggingFace_model_repos_internlm2-chat-7b
---------------------------------------------- --------- -------- ------ --------------------------------------------------------------------------
ceval-computer_network - - - -
ceval-operating_system - - - -
ceval-computer_architecture - - - -
ceval-college_programming 4ca32a accuracy gen 51.35
ceval-college_physics - - - -
ceval-college_chemistry e78857 accuracy gen 33.33
ceval-advanced_mathematics - - - -
ceval-probability_and_statistics - - - -
ceval-discrete_mathematics - - - -
ceval-electrical_engineer ae42b9 accuracy gen 40.54
ceval-metrology_engineer ee34ea accuracy gen 58.33
ceval-high_school_mathematics - - - -
ceval-high_school_physics - - - -
ceval-high_school_chemistry - - - -
ceval-high_school_biology - - - -
ceval-middle_school_mathematics - - - -
ceval-middle_school_biology - - - -
ceval-middle_school_physics - - - -
ceval-middle_school_chemistry - - - -
ceval-veterinary_medicine b4e08d accuracy gen 39.13
ceval-college_economics f3f4e6 accuracy gen 47.27
ceval-business_administration c1614e accuracy gen 51.52
ceval-marxism - - - -
ceval-mao_zedong_thought 51c7a4 accuracy gen 70.83
ceval-education_science 591fee accuracy gen 72.41
ceval-teacher_qualification 4e4ced accuracy gen 79.55
ceval-high_school_politics - - - -
ceval-high_school_geography - - - -
ceval-middle_school_politics - - - -
ceval-middle_school_geography - - - -
ceval-modern_chinese_history fc01af accuracy gen 65.22
ceval-ideological_and_moral_cultivation - - - -
ceval-logic - - - -
ceval-law a110a1 accuracy gen 41.67
ceval-chinese_language_and_literature 0f8b68 accuracy gen 56.52
ceval-art_studies 2a1300 accuracy gen 69.70
ceval-professional_tour_guide 4e673e accuracy gen 86.21
ceval-legal_professional ce8787 accuracy gen 43.48
ceval-high_school_chinese - - - -
ceval-high_school_history - - - -
ceval-middle_school_history - - - -
ceval-civil_servant 87d061 accuracy gen 55.32
ceval-sports_science - - - -
ceval-plant_protection - - - -
ceval-basic_medicine - - - -
ceval-clinical_medicine - - - -
ceval-urban_and_rural_planner 95b885 accuracy gen 58.70
ceval-accountant 002837 accuracy gen 44.90
ceval-fire_engineer bc23f5 accuracy gen 38.71
ceval-environmental_impact_assessment_engineer c64e2d accuracy gen 45.16
ceval-tax_accountant 3a5e3c accuracy gen 51.02
ceval-physician 6e277d accuracy gen 51.02
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
687
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
