llama大模型部署

看模型加载的参数设置.

import torch

# 初始化Half Tensor
h = torch.tensor([1.0,2.0,3.0], dtype=torch.half)
# h = torch.tensor([1.0,2.0,3.0], dtype=torch.float16) # 跟上面一行一样.

# 查看数据类型
print(h.dtype)
import accelerate
import bitsandbytes
from transformers import AutoTokenizer, AutoModelForCausalLM,TextIteratorStreamer
from transformers import AlbertTokenizer, AlbertModel
model = AlbertModel.from_pretrained('./albert',device_map='auto',torch_dtype=torch.float16,load_in_8bit=True,low_cpu_mem_usage=True)
# torch_dtype 模型本身的类型, 不写的话就自己根据权重文件查询出来.这个是权重文件本身决定的,一般在config.json里面
# load_in_8bit 会把模型转化为8bit类型.这个可以自己设置.

print(1)

• low_cpu_mem_usage algorithm:

  This is an experimental function that loads the model using ~1x model size CPU memory
  
    Here is how it works:
  
    1. save which state_dict keys we have
    2. drop state_dict before the model is created, since the latter takes 1x model size CPU memory
    3. after the model has been instantiated switch to the meta device all params/buffers that
    are going to be replaced from the loaded state_dict
    4. load state_dict 2nd time
    5. replace the params/buffers from the state_dict
  
    Currently, it can't handle deepspeed ZeRO stage 3 and ignores loading errors

这个算法low_cpu_mem 如果设置True

那么他会进行.

把权重字典的keys保存下来.

然后把state_dict删除.

初始化模型.把需要加载的参数位置放到meta device里面.

再加载state_dict

可以节省cpu内存. 小内存时候需要打开.