微调(fine-tuning)是一种基于预训练模型,通过少量的调整(fine-tune)来适应新的任务或数据的方法。
微调是在预训练模型的基础上,将模型中一些层的权重参数进行微调,以适应新的数据集或任务。
预训练模型部分已经在大规模数据上得到了训练,它们通常是较为通用且高性能的模型,因此可以很好地作为新任务的起点。微调可以加快模型的收敛速度,降低模型过拟合的风险,并在不消耗过多计算资源的情况下获取较好的模型性能。
在大模型的下游应用中,经常会用到两种微调模式:增量预训练 和 指令跟随 。
增量预训练
增量预训练是一种在已有预训练模型(比如:InternLM基座模型)的基础上,利用特定领域的数据进行进一步训练的方法。它的目的是在保持模型原有能力的同时,注入新的领域知识,进一步优化现有的预训练模型,从而提升模型在特定领域任务中的表现(比如:InternLM垂类基座模型)。增量预训练模型能够接受少量的新数据进行更新并适应新的任务,而不需要重新训练整个模型,这种方式可以很好地利用现有的预训练模型的知识,并在新数据上获得更好的性能。
指令跟随
指令跟随是指让模型根据用户输入的指令来执行相应的操作。模型通过对大量自然语言指令和相应操作的数据进行训练,学习如何将指令分解为具体的子任务,并选择合适的模块来执行这些任务(比如:InternLM垂类对话模型)。
环境配置
win下专用(linux跳过):
不建议使用win,有很多坑
git clone https://github.com/microsoft/DeepSpeed.git
Set-Item Env:\DS_BUILD_OPS 0
cd .\DeepSpeed\
python setup.py bdist_wheel
cd .\dist\
pip install .\deepspeed-0.14.5+681be6f5-py3-none-any.whl
原因参考Python|Windows 安装 DeepSpeed 安装方法及报错 Unable to pre-compile async_io 处理_deepspeed windows-CSDN博客
安装xtuner(linux)
git clone -b v0.1.23 https://github.com/InternLM/XTuner /root/InternLM/code/XTuner
cd /root/InternLM/code/XTuner
pip install -e '.[deepspeed]'
Windows安装移除相关路径信息
检测安装
xtuner version
建议安装transformers版本为4.40
无微调测试
import copy
import warnings
from dataclasses import asdict, dataclass
from typing import Callable, List, Optional
import streamlit as st
import torch
from torch import nn
from transformers.generation.utils import (LogitsProcessorList,
StoppingCriteriaList)
from transformers.utils import logging
from transformers import AutoTokenizer, AutoModelForCausalLM # isort: skip
logger = logging.get_logger(__name__)
model_name_or_path = "D:\AIcode\InternLM2\weight\internlm2-chat-1_8b"
@dataclass
class GenerationConfig:
# this config is used for chat to provide more diversity
max_length: int = 2048
top_p: float = 0.75
temperature: float = 0.1
do_sample: bool = True
repetition_penalty: float = 1.000
@torch.inference_mode()
def generate_interactive(
model,
tokenizer,
prompt,
generation_config: Optional[GenerationConfig] = None,
logits_processor: Optional[LogitsProcessorList] = None,
stopping_criteria: Optional[StoppingCriteriaList] = None,
prefix_allowed_tokens_fn: Optional[Callable[[int, torch.Tensor],
List[int]]] = None,
additional_eos_token_id: Optional[int] = None,
**kwargs,
):
inputs = tokenizer([prompt], padding=True, return_tensors='pt')
input_length = len(inputs['input_ids'][0])
for k, v in inputs.items():
inputs[k] = v.cuda()
input_ids = inputs['input_ids']
_, input_ids_seq_length = input_ids.shape[0], input_ids.shape[-1]
if generation_config is None:
generation_config = model.generation_config
generation_config = copy.deepcopy(generation_config)
model_kwargs = generation_config.update(**kwargs)
bos_token_id, eos_token_id = ( # noqa: F841 # pylint: disable=W0612
generation_config.bos_token_id,
generation_config.eos_token_id,
)
if isinstance(eos_token_id, int):
eos_token_id = [eos_token_id]
if additional_eos_token_id is not None:
eos_token_id.append(additional_eos_token_id)
has_default_max_length = kwargs.get(
'max_length') is None and generation_config.max_length is not None
if has_default_max_length and generation_config.max_new_tokens is None:
warnings.warn(
f"Using 'max_length''s default ({repr(generation_config.max_length)}) \
to control the generation length. "
'This behaviour is deprecated and will be removed from the \
config in v5 of Transformers -- we'
' recommend using `max_new_tokens` to control the maximum \
length of the generation.',
UserWarning,
)
elif generation_config.max_new_tokens is not None:
generation_config.max_length = generation_config.max_new_tokens + \
input_ids_seq_length
if not has_default_max_length:
logger.warn( # pylint: disable=W4902
f"Both 'max_new_tokens' (={generation_config.max_new_tokens}) "
f"and 'max_length'(={generation_config.max_length}) seem to "
"have been set. 'max_new_tokens' will take precedence. "
'Please refer to the documentation for more information. '
'(https://huggingface.co/docs/transformers/main/'
'en/main_classes/text_generation)',
UserWarning,
)
if input_ids_seq_length >= generation_config.max_length:
input_ids_string = 'input_ids'
logger.warning(
f"Input length of {input_ids_string} is {input_ids_seq_length}, "
f"but 'max_length' is set to {generation_config.max_length}. "
'This can lead to unexpected behavior. You should consider'
" increasing 'max_new_tokens'.")
# 2. Set generation parameters if not already defined
logits_processor = logits_processor if logits_processor is not None \
else LogitsProcessorList()
stopping_criteria = stopping_criteria if stopping_criteria is not None \
else StoppingCriteriaList()
logits_processor = model._get_logits_processor(
generation_config=generation_config,
input_ids_seq_length=input_ids_seq_length,
encoder_input_ids=input_ids,
prefix_allowed_tokens_fn=prefix_allowed_tokens_fn,
logits_processor=logits_processor,
)
stopping_criteria = model._get_stopping_criteria(
generation_config=generation_config,
stopping_criteria=stopping_criteria)
logits_warper = model._get_logits_warper(generation_config)
unfinished_sequences = input_ids.new(input_ids.shape[0]).fill_(1)
scores = None
while True:
model_inputs = model.prepare_inputs_for_generation(
input_ids, **model_kwargs)
# forward pass to get next token
outputs = model(
**model_inputs,
return_dict=True,
output_attentions=False,
output_hidden_states=False,
)
next_token_logits = outputs.logits[:, -1, :]
# pre-process distribution
next_token_scores = logits_processor(input_ids, next_token_logits)
next_token_scores = logits_warper(input_ids, next_token_scores)
# sample
probs = nn.functional.softmax(next_token_scores, dim=-1)
if generation_config.do_sample:
next_tokens = torch.multinomial(probs, num_samples=1).squeeze(1)
else:
next_tokens = torch.argmax(probs, dim=-1)
# update generated ids, model inputs, and length for next step
input_ids = torch.cat([input_ids, next_tokens[:, None]], dim=-1)
model_kwargs = model._update_model_kwargs_for_generation(
outputs, model_kwargs, is_encoder_decoder=False)
unfinished_sequences = unfinished_sequences.mul(
(min(next_tokens != i for i in eos_token_id)).long())
output_token_ids = input_ids[0].cpu().tolist()
output_token_ids = output_token_ids[input_length:]
for each_eos_token_id in eos_token_id:
if output_token_ids[-1] == each_eos_token_id:
output_token_ids = output_token_ids[:-1]
response = tokenizer.decode(output_token_ids)
yield response
# stop when each sentence is finished
# or if we exceed the maximum length
if unfinished_sequences.max() == 0 or stopping_criteria(
input_ids, scores):
break
def on_btn_click():
del st.session_state.messages
@st.cache_resource
def load_model():
model = (AutoModelForCausalLM.from_pretrained(model_name_or_path,
trust_remote_code=True).to(
torch.bfloat16).cuda())
tokenizer = AutoTokenizer.from_pretrained(model_name_or_path,
trust_remote_code=True)
return model, tokenizer
def prepare_generation_config():
with st.sidebar:
max_length = st.slider('Max Length',
min_value=8,
max_value=32768,
value=2048)
top_p = st.slider('Top P', 0.0, 1.0, 0.75, step=0.01)
temperature = st.slider('Temperature', 0.0, 1.0, 0.1, step=0.01)
st.button('Clear Chat History', on_click=on_btn_click)
generation_config = GenerationConfig(max_length=max_length,
top_p=top_p,
temperature=temperature)
return generation_config
user_prompt = '<|im_start|>user\n{user}<|im_end|>\n'
robot_prompt = '<|im_start|>assistant\n{robot}<|im_end|>\n'
cur_query_prompt = '<|im_start|>user\n{user}<|im_end|>\n\
<|im_start|>assistant\n'
def combine_history(prompt):
messages = st.session_state.messages
meta_instruction = ('')
total_prompt = f"<s><|im_start|>system\n{meta_instruction}<|im_end|>\n"
for message in messages:
cur_content = message['content']
if message['role'] == 'user':
cur_prompt = user_prompt.format(user=cur_content)
elif message['role'] == 'robot':
cur_prompt = robot_prompt.format(robot=cur_content)
else:
raise RuntimeError
total_prompt += cur_prompt
total_prompt = total_prompt + cur_query_prompt.format(user=prompt)
return total_prompt
def main():
# torch.cuda.empty_cache()
print('load model begin.')
model, tokenizer = load_model()
print('load model end.')
st.title('InternLM2-Chat-1.8B')
generation_config = prepare_generation_config()
# Initialize chat history
if 'messages' not in st.session_state:
st.session_state.messages = []
# Display chat messages from history on app rerun
for message in st.session_state.messages:
with st.chat_message(message['role'], avatar=message.get('avatar')):
st.markdown(message['content'])
# Accept user input
if prompt := st.chat_input('What is up?'):
# Display user message in chat message container
with st.chat_message('user'):
st.markdown(prompt)
real_prompt = combine_history(prompt)
# Add user message to chat history
st.session_state.messages.append({
'role': 'user',
'content': prompt,
})
with st.chat_message('robot'):
message_placeholder = st.empty()
for cur_response in generate_interactive(
model=model,
tokenizer=tokenizer,
prompt=real_prompt,
additional_eos_token_id=92542,
**asdict(generation_config),
):
# Display robot response in chat message container
message_placeholder.markdown(cur_response + '▌')
message_placeholder.markdown(cur_response)
# Add robot response to chat history
st.session_state.messages.append({
'role': 'robot',
'content': cur_response, # pylint: disable=undefined-loop-variable
})
torch.cuda.empty_cache()
if __name__ == '__main__':
main()
运行
streamlit run ./tool/xtuner_streamlit_demo.py
效果
生成训练数据
import json
# 设置用户的名字
#name = '伍鲜同志'
# 设置需要重复添加的数据次数
n = 3750
# 初始化数据
data = [
{"conversation": [{"input": "请介绍一下你自己", "output": "我是fengyanzi的小助手,内在是上海AI实验室书生·浦语的1.8B大模型哦"}]},
{"conversation": [{"input": "你在实战营做什么", "output": "我在这里帮助fengyanzi完成XTuner微调大模型测试"}]}
]
# 通过循环,将初始化的对话数据重复添加到data列表中
for i in range(n):
data.append(data[0])
data.append(data[1])
# 将data列表中的数据写入到'datas/assistant.json'文件中
with open('/root/InternLM/XTuner/datas/assistant.json', 'w', encoding='utf-8') as f:
# 使用json.dump方法将数据以JSON格式写入文件
# ensure_ascii=False 确保中文字符正常显示
# indent=4 使得文件内容格式化,便于阅读
json.dump(data, f, ensure_ascii=False, indent=4)
设置配置文件
xtuner list-cfg -p internlm2
使用预设训练参数
xtuner copy-cfg internlm2_chat_1_8b_qlora_alpaca_e3 .
修改设置
#######################################################################
# PART 1 Settings #
#######################################################################
- pretrained_model_name_or_path = 'internlm/internlm2-chat-1_8b'
+ pretrained_model_name_or_path = '/root/InternLM/XTuner/Shanghai_AI_Laboratory/internlm2-chat-1_8b'
- alpaca_en_path = 'tatsu-lab/alpaca'
+ alpaca_en_path = 'datas/assistant.json'
evaluation_inputs = [
- '请给我介绍五个上海的景点', 'Please tell me five scenic spots in Shanghai'
+ '请介绍一下你自己', 'Please introduce yourself'
]
#######################################################################
# PART 3 Dataset & Dataloader #
#######################################################################
alpaca_en = dict(
type=process_hf_dataset,
- dataset=dict(type=load_dataset, path=alpaca_en_path),
+ dataset=dict(type=load_dataset, path='json', data_files=dict(train=alpaca_en_path)),
tokenizer=tokenizer,
max_length=max_length,
- dataset_map_fn=alpaca_map_fn,
+ dataset_map_fn=None,
template_map_fn=dict(
type=template_map_fn_factory, template=prompt_template),
remove_unused_columns=True,
shuffle_before_pack=True,
pack_to_max_length=pack_to_max_length,
use_varlen_attn=use_varlen_attn)
启动微调
xtuner train ./internlm2_chat_1_8b_qlora_alpaca_e3_copy.py
最汗流浃背的一集
等待两分钟后加载完成开始训练,1.8B模型推理消耗4G显存,Qlora微调消耗8GB显存
模型转换
cd /root/InternLM/XTuner
conda activate xtuner0121
# 先获取最后保存的一个pth文件
pth_file=`ls -t ./work_dirs/internlm2_chat_1_8b_qlora_alpaca_e3_copy/*.pth | head -n 1`
export MKL_SERVICE_FORCE_INTEL=1
export MKL_THREADING_LAYER=GNU
xtuner convert pth_to_hf ./internlm2_chat_1_8b_qlora_alpaca_e3_copy.py ${pth_file} ./hf
模型合并
export MKL_SERVICE_FORCE_INTEL=1
export MKL_THREADING_LAYER=GNU
xtuner convert merge /root/InternLM/XTuner/Shanghai_AI_Laboratory/internlm2-chat-1_8b ./hf ./merged --max-shard-size 2GB
测试
# 直接修改脚本文件第18行
- model_name_or_path = "/root/InternLM/XTuner/Shanghai_AI_Laboratory/internlm2-chat-1_8b"
+ model_name_or_path = "/root/InternLM/XTuner/merged"
streamlit run /root/InternLM/Tutorial/tools/xtuner_streamlit_demo.py
权重地址:模型中心-OpenXLab