书生浦语大模型实战营第六次课作业-OpenCompass大模型评测

基础作业

• 使用 OpenCompass 评测 InternLM2-Chat-7B 模型在 C-Eval 数据集上的性能

进阶作业

• 使用 OpenCompass 评测 InternLM2-Chat-7B 模型使用 LMDeploy 0.2.0 部署后在 C-Eval 数据集上的性能

1. 基础作业

使用 OpenCompass 评测 InternLM2-Chat-7B 模型在 C-Eval 数据集上的性能

1.1 修改配置文件

from opencompass.models import HuggingFaceCausalLM
from mmengine.config import read_base

with read_base():
    from .datasets.ceval.ceval_gen_5f30c7 import ceval_datasets  # noqa: F401, F403
    # and output the results in a choosen format
    #from .summarizers.medium import summarizer


_meta_template = dict(
    round=[
        dict(role='HUMAN', begin='<|im_start|>user\n', end='<|im_end|>\n'),
        dict(role='SYSTEM', begin='<|im_start|>system\n', end='<|im_end|>\n'),
        dict(role='BOT', begin='<|im_start|>assistant\n', end='<|im_end|>\n', generate=True),
    ],
    eos_token_id=92542
)

internlm_meta_template = dict(round=[
    dict(role='HUMAN', begin='<|User|>:', end='\n'),
    dict(role='BOT', begin='<|Bot|>:', end='<eoa>\n', generate=True),
],
                              eos_token_id=103028)

models = [
    dict(
        type=HuggingFaceCausalLM,
        abbr='internlm2-chat-7b',
        path="/root/share/model_repos/internlm2-chat-7b/",
        tokenizer_path='/root/share/model_repos/internlm2-chat-7b/',
        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=16,
        max_seq_len=2048,
        batch_size=4,
        meta_template=internlm_meta_template,
        run_cfg=dict(num_gpus=1, num_procs=1),
        # end_str='<|im_end|>',
    )
]

datasets = [*ceval_datasets]

1.2 启动评测

python run.py configs/eval_internlm2_chat_7b.py

1.3 评测完成

2. 进阶作业

使用 OpenCompass 评测 InternLM2-Chat-7B 模型使用 LMDeploy 0.2.0 部署后在 C-Eval 数据集上的性能

参考文档:

评测 LMDeploy 模型 — OpenCompass 0.2.1 文档

2.1 安装lmdeploy0.2.0

2.2 修改配置文件

from mmengine.config import read_base
from opencompass.models.turbomind import TurboMindModel

with read_base():
    # choose a list of datasets
    # from .datasets.mmlu.mmlu_gen_a484b3 import mmlu_datasets
    from .datasets.ceval.ceval_gen_5f30c7 import ceval_datasets
    # from .datasets.SuperGLUE_WiC.SuperGLUE_WiC_gen_d06864 import WiC_datasets
    # from .datasets.SuperGLUE_WSC.SuperGLUE_WSC_gen_7902a7 import WSC_datasets
    # from .datasets.triviaqa.triviaqa_gen_2121ce import triviaqa_datasets
    # from .datasets.gsm8k.gsm8k_gen_1d7fe4 import gsm8k_datasets
    # from .datasets.race.race_gen_69ee4f import race_datasets
    # from .datasets.crowspairs.crowspairs_gen_381af0 import crowspairs_datasets
    # and output the results in a choosen format
    from .summarizers.medium import summarizer


datasets = sum((v for k, v in locals().items() if k.endswith('_datasets')), [])

internlm_meta_template = dict(round=[
    dict(role='HUMAN', begin='<|User|>:', end='\n'),
    dict(role='BOT', begin='<|Bot|>:', end='<eoa>\n', generate=True),
],
                              eos_token_id=103028)

# config for internlm2-chat-7b
internlm2_chat_7b = dict(
    type=TurboMindModel,
    abbr='internlm2-chat-7b-turbomind',
    path='/root/share/model_repos/internlm2-chat-7b/',
    engine_config=dict(session_len=2048,
                       max_batch_size=32,
                       rope_scaling_factor=1.0),
    gen_config=dict(top_k=1,
                    top_p=0.8,
                    temperature=1.0,
                    max_new_tokens=100),
    max_out_len=16,
    max_seq_len=2048,
    batch_size=4,
    concurrency=32,
    meta_template=internlm_meta_template,
    run_cfg=dict(num_gpus=1, num_procs=1),
)


models = [internlm2_chat_7b]

2.3 启动评测

python run.py configs/eval_internlm2_chat_7b_turbomind.py 

若报错显存不足，换成显存40G的开发机即可

​​​​​​​2.4 评测完成

2.5 对比

使用LMDepoly平均分：59.37

未使用LMDepoly平均分：60.62

使用LMDeploy花费时间：

未使用LMDeploy花费时间：