[Codellama assessment score with meta prompt is half lower than codellama assessment score without meta prompt]

Prerequisite

• I have searched Issues and Discussions but cannot get the expected help.
• The bug has not been fixed in the latest version.

Type

I'm evaluating with the officially supported tasks/models/datasets.

Environment

{'CUDA available': True,
 'CUDA_HOME': '/usr/local/cuda-12.2',
 'GCC': 'gcc (GCC) 7.3.0',
 'GPU 0,1,2,3,4,5,6,7': 'NVIDIA A100-SXM4-40GB',
 'MMEngine': '0.10.1',
 'NVCC': 'Cuda compilation tools, release 12.2, V12.2.128',
 'OpenCV': '4.8.1',
 'PyTorch': '2.1.2',
 'PyTorch compiling details': 'PyTorch built with:\n'
                              '  - GCC 9.3\n'
                              '  - C++ Version: 201703\n'
                              '  - Intel(R) oneAPI Math Kernel Library Version '
                              '2023.1-Product Build 20230303 for Intel(R) 64 '
                              'architecture applications\n'
                              '  - Intel(R) MKL-DNN v3.1.1 (Git Hash '
                              '64f6bcbcbab628e96f33a62c3e975f8535a7bde4)\n'
                              '  - OpenMP 201511 (a.k.a. OpenMP 4.5)\n'
                              '  - LAPACK is enabled (usually provided by '
                              'MKL)\n'
                              '  - NNPACK is enabled\n'
                              '  - CPU capability usage: AVX2\n'
                              '  - CUDA Runtime 12.1\n'
                              '  - NVCC architecture flags: '
                              '-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_61,code=sm_61;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90\n'
                              '  - CuDNN 8.9.2\n'
                              '  - Magma 2.6.1\n'
                              '  - Build settings: BLAS_INFO=mkl, '
                              'BUILD_TYPE=Release, CUDA_VERSION=12.1, '
                              'CUDNN_VERSION=8.9.2, '
                              'CXX_COMPILER=/opt/rh/devtoolset-9/root/usr/bin/c++, '
                              'CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=0 '
                              '-fabi-version=11 -fvisibility-inlines-hidden '
                              '-DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO '
                              '-DLIBKINETO_NOROCTRACER -DUSE_FBGEMM '
                              '-DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK '
                              '-DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE '
                              '-O2 -fPIC -Wall -Wextra -Werror=return-type '
                              '-Werror=non-virtual-dtor -Werror=bool-operation '
                              '-Wnarrowing -Wno-missing-field-initializers '
                              '-Wno-type-limits -Wno-array-bounds '
                              '-Wno-unknown-pragmas -Wno-unused-parameter '
                              '-Wno-unused-function -Wno-unused-result '
                              '-Wno-strict-overflow -Wno-strict-aliasing '
                              '-Wno-stringop-overflow -Wno-psabi '
                              '-Wno-error=pedantic -Wno-error=old-style-cast '
                              '-Wno-invalid-partial-specialization '
                              '-Wno-unused-private-field '
                              '-Wno-aligned-allocation-unavailable '
                              '-Wno-missing-braces -fdiagnostics-color=always '
                              '-faligned-new -Wno-unused-but-set-variable '
                              '-Wno-maybe-uninitialized -fno-math-errno '
                              '-fno-trapping-math -Werror=format '
                              '-Werror=cast-function-type '
                              '-Wno-stringop-overflow, LAPACK_INFO=mkl, '
                              'PERF_WITH_AVX=1, PERF_WITH_AVX2=1, '
                              'PERF_WITH_AVX512=1, '
                              'TORCH_DISABLE_GPU_ASSERTS=ON, '
                              'TORCH_VERSION=2.1.2, USE_CUDA=ON, USE_CUDNN=ON, '
                              'USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, '
                              'USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, '
                              'USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, '
                              'USE_OPENMP=ON, USE_ROCM=OFF, \n',
 'Python': '3.10.13 (main, Sep 11 2023, 13:44:35) [GCC 11.2.0]',
 'TorchVision': '0.16.2',
 'numpy_random_seed': 2147483648,
 'opencompass': '0.2.0+c3e0fcf',
 'sys.platform': 'linux'}

Reproduces the problem - code/configuration sample

The code of Codellama assessment process with meta prompt is as follows:

#eval_codellama_with_metaprompt.py
from mmengine.config import read_base
from opencompass.models import HuggingFaceCausalLM

with read_base():
    from .datasets.humaneval.humaneval_gen_8e312c import humaneval_datasets  # noqa: F401, F403

llama2_meta_template = dict(
    round=[
        dict(role='HUMAN', begin='<s>[INST] ', end=' [/INST] '),
        dict(role='BOT', begin='', end=' </s>', generate=True),
    ],
    eos_token_id=2)

models = [
    dict(
        type=HuggingFaceCausalLM,
        abbr='CodeLlama-13b-Instruct-hf',
        path="the local path of CodeLlama-13b-Instruct-hf",
        tokenizer_path='the local path of CodeLlama-13b-Instruct-hf',
        tokenizer_kwargs=dict(padding_side='left',
                              truncation_side='left',
                              use_fast=False,
                              ),
        max_out_len=100,
        max_seq_len=2048,
        batch_size=8,
        model_kwargs=dict(device_map='auto'),
        batch_padding=False,  # if false, inference with for-loop without batch padding
        meta_template=llama2_meta_template,
        run_cfg=dict(num_gpus=8, num_procs=1),
    )
]

datasets = [*humaneval_datasets]

The code of Codellama assessment process without meta prompt is as follows:

#eval_codellama_without_metaprompt.py
from mmengine.config import read_base
from opencompass.models import HuggingFaceCausalLM

with read_base():
    from .datasets.humaneval.humaneval_gen_8e312c import humaneval_datasets  # noqa: F401, F403

models = [
    dict(
        type=HuggingFaceCausalLM,
        abbr='CodeLlama-13b-Instruct-hf',
        path="the local path of CodeLlama-13b-Instruct-hf",
        tokenizer_path='the local path of CodeLlama-13b-Instruct-hf',
        tokenizer_kwargs=dict(padding_side='left',
                              truncation_side='left',
                              use_fast=False,
                              ),
        max_out_len=100,
        max_seq_len=2048,
        batch_size=8,
        model_kwargs=dict(device_map='auto'),
        batch_padding=False,  # if false, inference with for-loop without batch padding
        run_cfg=dict(num_gpus=8, num_procs=1),
    )
]

datasets = [*humaneval_datasets]

Reproduces the problem - command or script

python run.py configs/eval_codellama_with_metaprompt.py
python run.py configs/eval_codellama_without_metaprompt.py

Reproduces the problem - error message

eval_codellama_with_metaprompt.py

dataset           version    metric            mode      CodeLlama-13b-Instruct-hf
----------------  ---------  ----------------  ------  ---------------------------
openai_humaneval  8e312c     humaneval_pass@1  gen                            12.8

eval_codellama_without_metaprompt.py

dataset           version    metric            mode      CodeLlama-13b-Instruct-hf
----------------  ---------  ----------------  ------  ---------------------------
openai_humaneval  8e312c     humaneval_pass@1  gen                            28.05

Other information

As the website say: https://opencompass.readthedocs.io/en/latest/prompt/meta_template.html
I think codellama assessment score with meta prompt can be high than codellama assessment score without meta prompt