使用Accelerate库在多GPU上进行LLM推理

2023-11-29 12:28

本文主要是介绍使用Accelerate库在多GPU上进行LLM推理,希望对大家解决编程问题提供一定的参考价值,需要的开发者们随着小编来一起学习吧!

大型语言模型(llm)已经彻底改变了自然语言处理领域。随着这些模型在规模和复杂性上的增长,推理的计算需求也显著增加。为了应对这一挑战利用多个gpu变得至关重要。

所以本文将在多个gpu上并行执行推理,主要包括:Accelerate库介绍,简单的方法与工作代码示例和使用多个gpu的性能基准测试。

本文将使用多个3090将llama2-7b的推理扩展在多个GPU上

基本示例

我们首先介绍一个简单的示例来演示使用Accelerate进行多gpu“消息传递”。

 from accelerate import Acceleratorfrom accelerate.utils import gather_objectaccelerator = Accelerator()# each GPU creates a stringmessage=[ f"Hello this is GPU {accelerator.process_index}" ] # collect the messages from all GPUsmessages=gather_object(message)# output the messages only on the main process with accelerator.print() accelerator.print(messages)

输出如下:

 ['Hello this is GPU 0', 'Hello this is GPU 1', 'Hello this is GPU 2', 'Hello this is GPU 3', 'Hello this is GPU 4']

多GPU推理

下面是一个简单的、非批处理的推理方法。代码很简单,因为Accelerate库已经帮我们做了很多工作,我们直接使用就可以:

 from accelerate import Acceleratorfrom accelerate.utils import gather_objectfrom transformers import AutoModelForCausalLM, AutoTokenizerfrom statistics import meanimport torch, time, jsonaccelerator = Accelerator()# 10*10 Prompts. Source: https://www.penguin.co.uk/articles/2022/04/best-first-lines-in-booksprompts_all=["The King is dead. Long live the Queen.","Once there were four children whose names were Peter, Susan, Edmund, and Lucy.","The story so far: in the beginning, the universe was created.","It was a bright cold day in April, and the clocks were striking thirteen.","It is a truth universally acknowledged, that a single man in possession of a good fortune, must be in want of a wife.","The sweat wis lashing oafay Sick Boy; he wis trembling.","124 was spiteful. Full of Baby's venom.","As Gregor Samsa awoke one morning from uneasy dreams he found himself transformed in his bed into a gigantic insect.","I write this sitting in the kitchen sink.","We were somewhere around Barstow on the edge of the desert when the drugs began to take hold.",] * 10# load a base model and tokenizermodel_path="models/llama2-7b"model = AutoModelForCausalLM.from_pretrained(model_path,    device_map={"": accelerator.process_index},torch_dtype=torch.bfloat16,)tokenizer = AutoTokenizer.from_pretrained(model_path)   # sync GPUs and start the timeraccelerator.wait_for_everyone()start=time.time()# divide the prompt list onto the available GPUs with accelerator.split_between_processes(prompts_all) as prompts:# store output of generations in dictresults=dict(outputs=[], num_tokens=0)# have each GPU do inference, prompt by promptfor prompt in prompts:prompt_tokenized=tokenizer(prompt, return_tensors="pt").to("cuda")output_tokenized = model.generate(**prompt_tokenized, max_new_tokens=100)[0]# remove prompt from output output_tokenized=output_tokenized[len(prompt_tokenized["input_ids"][0]):]# store outputs and number of tokens in result{}results["outputs"].append( tokenizer.decode(output_tokenized) )results["num_tokens"] += len(output_tokenized)results=[ results ] # transform to list, otherwise gather_object() will not collect correctly# collect results from all the GPUsresults_gathered=gather_object(results)if accelerator.is_main_process:timediff=time.time()-startnum_tokens=sum([r["num_tokens"] for r in results_gathered ])print(f"tokens/sec: {num_tokens//timediff}, time {timediff}, total tokens {num_tokens}, total prompts {len(prompts_all)}")

使用多个gpu会导致一些通信开销:性能在4个gpu时呈线性增长,然后在这种特定设置中趋于稳定。当然这里的性能取决于许多参数,如模型大小和量化、提示长度、生成的令牌数量和采样策略,所以我们只讨论一般的情况

1 GPU: 44个token /秒,时间:225.5s

2 gpu: 88个token /秒,时间:112.9s

3 gpu: 128个token /秒,时间:77.6s

4 gpu: 137个token /秒,时间:72.7s

5 gpu: 119个token /秒,时间:83.8s

在多GPU上进行批处理

现实世界中,我们可以使用批处理推理来加快速度。这会减少GPU之间的通讯,加快推理速度。我们只需要增加prepare_prompts函数将一批数据而不是单条数据输入到模型即可:

 from accelerate import Acceleratorfrom accelerate.utils import gather_objectfrom transformers import AutoModelForCausalLM, AutoTokenizerfrom statistics import meanimport torch, time, jsonaccelerator = Accelerator()def write_pretty_json(file_path, data):import jsonwith open(file_path, "w") as write_file:json.dump(data, write_file, indent=4)# 10*10 Prompts. Source: https://www.penguin.co.uk/articles/2022/04/best-first-lines-in-booksprompts_all=["The King is dead. Long live the Queen.","Once there were four children whose names were Peter, Susan, Edmund, and Lucy.","The story so far: in the beginning, the universe was created.","It was a bright cold day in April, and the clocks were striking thirteen.","It is a truth universally acknowledged, that a single man in possession of a good fortune, must be in want of a wife.","The sweat wis lashing oafay Sick Boy; he wis trembling.","124 was spiteful. Full of Baby's venom.","As Gregor Samsa awoke one morning from uneasy dreams he found himself transformed in his bed into a gigantic insect.","I write this sitting in the kitchen sink.","We were somewhere around Barstow on the edge of the desert when the drugs began to take hold.",] * 10# load a base model and tokenizermodel_path="models/llama2-7b"model = AutoModelForCausalLM.from_pretrained(model_path,    device_map={"": accelerator.process_index},torch_dtype=torch.bfloat16,)tokenizer = AutoTokenizer.from_pretrained(model_path)   tokenizer.pad_token = tokenizer.eos_token# batch, left pad (for inference), and tokenizedef prepare_prompts(prompts, tokenizer, batch_size=16):batches=[prompts[i:i + batch_size] for i in range(0, len(prompts), batch_size)]  batches_tok=[]tokenizer.padding_side="left"     for prompt_batch in batches:batches_tok.append(tokenizer(prompt_batch, return_tensors="pt", padding='longest', truncation=False, pad_to_multiple_of=8,add_special_tokens=False).to("cuda") )tokenizer.padding_side="right"return batches_tok# sync GPUs and start the timeraccelerator.wait_for_everyone()    start=time.time()# divide the prompt list onto the available GPUs with accelerator.split_between_processes(prompts_all) as prompts:results=dict(outputs=[], num_tokens=0)# have each GPU do inference in batchesprompt_batches=prepare_prompts(prompts, tokenizer, batch_size=16)for prompts_tokenized in prompt_batches:outputs_tokenized=model.generate(**prompts_tokenized, max_new_tokens=100)# remove prompt from gen. tokensoutputs_tokenized=[ tok_out[len(tok_in):] for tok_in, tok_out in zip(prompts_tokenized["input_ids"], outputs_tokenized) ] # count and decode gen. tokens num_tokens=sum([ len(t) for t in outputs_tokenized ])outputs=tokenizer.batch_decode(outputs_tokenized)# store in results{} to be gathered by accelerateresults["outputs"].extend(outputs)results["num_tokens"] += num_tokensresults=[ results ] # transform to list, otherwise gather_object() will not collect correctly# collect results from all the GPUsresults_gathered=gather_object(results)if accelerator.is_main_process:timediff=time.time()-startnum_tokens=sum([r["num_tokens"] for r in results_gathered ])print(f"tokens/sec: {num_tokens//timediff}, time elapsed: {timediff}, num_tokens {num_tokens}")

可以看到批处理会大大加快速度。

1 GPU: 520 token /sec,时间:19.2s

2 gpu: 900 token /sec,时间:11.1s

3 gpu: 1205个token /秒,时间:8.2s

4 gpu: 1655 token /sec,时间:6.0s

5 gpu: 1658 token /sec,时间:6.0s

总结

截止到本文为止,llama.cpp,ctransformer还不支持多GPU推理,好像llama.cpp在6月有个多GPU的merge,但是我没看到官方更新,所以这里暂时确定不支持多GPU。如果有小伙伴确认可以支持多GPU请留言。

huggingface的Accelerate包则为我们使用多GPU提供了一个很方便的选择,使用多个GPU推理可以显着提高性能,但gpu之间通信的开销随着gpu数量的增加而显著增加。

https://avoid.overfit.cn/post/8210f640cae0404a88fd1c9028c6aabb

作者:Geronimo

这篇关于使用Accelerate库在多GPU上进行LLM推理的文章就介绍到这儿,希望我们推荐的文章对编程师们有所帮助!



http://www.chinasem.cn/article/432819

相关文章

一文详解如何使用Java获取PDF页面信息

《一文详解如何使用Java获取PDF页面信息》了解PDF页面属性是我们在处理文档、内容提取、打印设置或页面重组等任务时不可或缺的一环,下面我们就来看看如何使用Java语言获取这些信息吧... 目录引言一、安装和引入PDF处理库引入依赖二、获取 PDF 页数三、获取页面尺寸(宽高)四、获取页面旋转角度五、判断

C++中assign函数的使用

《C++中assign函数的使用》在C++标准模板库中,std::list等容器都提供了assign成员函数,它比操作符更灵活,支持多种初始化方式,下面就来介绍一下assign的用法,具有一定的参考价... 目录​1.assign的基本功能​​语法​2. 具体用法示例​​​(1) 填充n个相同值​​(2)

MySQL深分页进行性能优化的常见方法

《MySQL深分页进行性能优化的常见方法》在Web应用中,分页查询是数据库操作中的常见需求,然而,在面对大型数据集时,深分页(deeppagination)却成为了性能优化的一个挑战,在本文中,我们将... 目录引言:深分页,真的只是“翻页慢”那么简单吗?一、背景介绍二、深分页的性能问题三、业务场景分析四、

Spring StateMachine实现状态机使用示例详解

《SpringStateMachine实现状态机使用示例详解》本文介绍SpringStateMachine实现状态机的步骤,包括依赖导入、枚举定义、状态转移规则配置、上下文管理及服务调用示例,重点解... 目录什么是状态机使用示例什么是状态机状态机是计算机科学中的​​核心建模工具​​,用于描述对象在其生命

使用Python删除Excel中的行列和单元格示例详解

《使用Python删除Excel中的行列和单元格示例详解》在处理Excel数据时,删除不需要的行、列或单元格是一项常见且必要的操作,本文将使用Python脚本实现对Excel表格的高效自动化处理,感兴... 目录开发环境准备使用 python 删除 Excphpel 表格中的行删除特定行删除空白行删除含指定

SpringBoot结合Docker进行容器化处理指南

《SpringBoot结合Docker进行容器化处理指南》在当今快速发展的软件工程领域,SpringBoot和Docker已经成为现代Java开发者的必备工具,本文将深入讲解如何将一个SpringBo... 目录前言一、为什么选择 Spring Bootjavascript + docker1. 快速部署与

深入理解Go语言中二维切片的使用

《深入理解Go语言中二维切片的使用》本文深入讲解了Go语言中二维切片的概念与应用,用于表示矩阵、表格等二维数据结构,文中通过示例代码介绍的非常详细,需要的朋友们下面随着小编来一起学习学习吧... 目录引言二维切片的基本概念定义创建二维切片二维切片的操作访问元素修改元素遍历二维切片二维切片的动态调整追加行动态

prometheus如何使用pushgateway监控网路丢包

《prometheus如何使用pushgateway监控网路丢包》:本文主要介绍prometheus如何使用pushgateway监控网路丢包问题,具有很好的参考价值,希望对大家有所帮助,如有错误... 目录监控网路丢包脚本数据图表总结监控网路丢包脚本[root@gtcq-gt-monitor-prome

Python通用唯一标识符模块uuid使用案例详解

《Python通用唯一标识符模块uuid使用案例详解》Pythonuuid模块用于生成128位全局唯一标识符,支持UUID1-5版本,适用于分布式系统、数据库主键等场景,需注意隐私、碰撞概率及存储优... 目录简介核心功能1. UUID版本2. UUID属性3. 命名空间使用场景1. 生成唯一标识符2. 数

linux解压缩 xxx.jar文件进行内部操作过程

《linux解压缩xxx.jar文件进行内部操作过程》:本文主要介绍linux解压缩xxx.jar文件进行内部操作,具有很好的参考价值,希望对大家有所帮助,如有错误或未考虑完全的地方,望不吝赐教... 目录一、解压文件二、压缩文件总结一、解压文件1、把 xxx.jar 文件放在服务器上,并进入当前目录#