修改配置文件，无法测试本地模型，没有报错信息[Bug] #811
Replies: 11 comments 3 replies
-
|
麻烦帮忙看下，本地已有模型，修改了 path 和 tokenizer_path，但是运行不起来。 |
Beta Was this translation helpful? Give feedback.
-
|
plz show the content of "D:\code\ChatGLM-6B-main\THUDM\chatglm2-6". |
Beta Was this translation helpful? Give feedback.
-
|
|
Beta Was this translation helpful? Give feedback.
-
这个文件内容为空。 |
Beta Was this translation helpful? Give feedback.
-
|
Beta Was this translation helpful? Give feedback.
-
|
你可以在运行的时候，加上 --debug，看下报错信息。比如python run.py ChatGLM2-6B.py --debug |
Beta Was this translation helpful? Give feedback.
-
debug 模式下并没有发现错误，但是最后没有分数，debug 内容如下： ceval-computer_network - - - - (opencompass0112) D:\code\opencompass> |
Beta Was this translation helpful? Give feedback.
-
|
为什么 datasets = [*siqa_datasets]，log是ceval? 提供一个完整的config文件吧 |
Beta Was this translation helpful? Give feedback.
-
from mmengine.config import read_base with read_base(): datasets = [*ceval_datasets] from opencompass.models import HuggingFaceCausalLM models = [ |
Beta Was this translation helpful? Give feedback.
-
|
why use num_gpus=0? Please try num_gpus=1 |
Beta Was this translation helpful? Give feedback.
-
|
我在使用opencompass的过程中同样遇到这个问题请问您是如何修复的 |
Beta Was this translation helpful? Give feedback.
-
|
+1，如果本地已经下载好了模型的huggingface版本的文件，怎么修改config进行加载能正常运行测试，看起来只修改path看起来是不work的 |
Beta Was this translation helpful? Give feedback.
-
|
qi
你好，我也遇到了类似问题，请问你解决了吗？ |
Beta Was this translation helpful? Give feedback.
-
你好，我也遇到了同样的问题，进度一直是0%，模型和数据集都是本地的 |
Beta Was this translation helpful? Give feedback.
-
先决条件
问题类型
我正在使用官方支持的任务/模型/数据集进行评估。
环境
‘’‘
python run.py configs\eval_chatglm6b.py
’‘’
重现问题 - 代码/配置示例
配置文件如下所示。
‘’‘
from mmengine.config import read_base
with read_base():
from .datasets.siqa.siqa_gen import siqa_datasets
datasets = [*siqa_datasets]
from opencompass.models import HuggingFaceCausalLM
models = [
dict(
type=HuggingFaceCausalLM,
abbr='chatglm6b', # 运行完结果展示的名称
path="D:\code\ChatGLM-6B-main\THUDM\chatglm2-6", # 模型路径
tokenizer_path='D:\code\ChatGLM-6B-main\THUDM\chatglm2-6', # 分词器路径
model_kwargs=dict(
device_map='auto',
trust_remote_code=True
),
tokenizer_kwargs=dict(
padding_side='left',
truncation_side='left',
trust_remote_code=True,
use_fast=False
),
pad_token_id=151643,
max_out_len=100,
max_seq_len=2048,
batch_size=16,
#run_cfg=dict(num_gpus=8, num_procs=8), # 我使用A800 开启8张卡评测，就算使用1，也默认使用多卡
)
]
’‘’
重现问题 - 命令或脚本
‘’‘
python run.py configs\eval_chatglm6b.py
’‘’
重现问题 - 错误信息
.out 文件没有错误信息
其他信息
No response
Beta Was this translation helpful? Give feedback.
All reactions