书生浦语的第二次作业，demo运行

wyx_buaa

已于 2024-04-17 15:13:28 修改

阅读量105

点赞数 3

文章标签： python numpy scipy pandas

于 2024-04-15 11:04:10 首次发布

本文链接：https://blog.csdn.net/qq_52233175/article/details/137773110

版权

本文介绍了如何使用Python脚本dl.py从Shanghai_AI_Laboratory下载InternLM-20b模型的config.json文件，并展示了配置文件的内容，包括模型架构、参数设置等详细信息。

摘要由CSDN通过智能技术生成

interlm-20b 的config.json 下载截图如如下，首先写入代码，文件名为dl.py

import torch
from modelscope import snapshot_download, AutoModel, AutoTokenizer
import os
model_dir = snapshot_download('Shanghai_AI_Laboratory/internlm-20b', cache_dir='/home/JJ_Group/wuyx/llm/', revision='master')

运行代码，截图如下

config.json如下：

{
"architectures": [
    "InternLMForCausalLM"
],
"auto_map": {
    "AutoConfig": "configuration_internlm.InternLMConfig",
    "AutoModel": "modeling_internlm.InternLMForCausalLM",
    "AutoModelForCausalLM": "modeling_internlm.InternLMForCausalLM"
},
"bias": false,
"bos_token_id": 1,
"eos_token_id": 2,
"hidden_act": "silu",
"hidden_size": 5120,
"initializer_range": 0.02,
"intermediate_size": 13824,
"max_position_embeddings": 4096,
"model_type": "internlm",
"num_attention_heads": 40,
"num_hidden_layers": 60,
"num_key_value_heads": 40,
"pad_token_id": 2,
"pretraining_tp": 1,
"rms_norm_eps": 1e-06,
"tie_word_embeddings": false,
"torch_dtype": "bfloat16",
"transformers_version": "4.33.1",
"use_cache": true,
"vocab_size": 103168,
"rotary": {
    "base": 10000,
    "type": "dynamic"
}
}