当前位置:首页 > 报告详情

使用开源 AI 进行构建:速成课程.pdf

上传人: 卢*** 编号:908333 2025-09-07 52页 10.94MB

1、A Crash Course:World Summit AIBuilding with Open Source AICedric ClyburnSenior Developer AdvocateRed Hat AI BUcedricclyburn1Legare KerrisonDeveloper AdvocateRed Hat AI BUlegarekerrisonopen sourcecloud computingand automationartificial intelligence and machine learningJan 2023:The ProblemFor 2023 and

2、 the start of 2024,closed dramatically outpaced open.Slide from Neural Magic Board Meeting in March 2023 The Power of OpenThere has been an explosion of capability from open-source over the last 2 years.Llama No OSS modelsZephyrLlama 2Mistral,Granite 2DeepSeek-R1Mixtral,Phi-2Jan 2023RedPajama,MPT,Fa

3、lconMar 2023May 2023July 2023Sept 2023Nov 2023Jan2024July2024Sept 2024Nov 2024Gemma2,NemotronJan 2025DBRX,Granite 3Qwen2-VLPhi-3,ArcticDBRX,Phi-3Llama 3,Qwen2Mar2024May2024The Power of OpenOpen models are deployment targets today.And the trend is not slowing down.650M downloads in 202485,000 Llama d

4、erivative models1B,3B,8B,70B,405B variantsMultilingual,Multimodal,MobileFirst reasoning model on par in quality with OpenAI O11-70B parameter distilled versionsGlobal market pandemonium?LlamaR1HeadlinesModels are commoditizing many options for diverse enterprise needs.Self managed infrastructure1B-4

5、05B size-match task difficulty to model CostComplete data privacy(no 3rd party APIs)SecurityModel lifecycle(no changes to the model in place)Resources(no rate limits/API downtime)ControlImprove accuracy and costs with task specific tuningCustomizationAdvantages of Open Source ModelsOpen-source model

6、s play an important role in the Enterprise AI landscape.7Open source is about more than developing software.Its how we built Red Hat.And its completely revolutionized the AI ecosystem too.Open source is great driver for InnovationThere are various organizations that define open source,but generallyD

word格式文档无特别注明外均可编辑修改,预览文件经过压缩,下载原文更清晰!
三个皮匠报告文库所有资源均是客户上传分享,仅供网友学习交流,未经上传用户书面授权,请勿作商用。
根据报告的内容,全文主要内容概括如下: 1. **开源AI的兴起**:过去两年中,开源AI模型能力爆炸式增长,如Llama、Mistral等,模型下载量巨大,衍生模型众多。 2. **开源模型优势**:开源模型在企业和AI生态系统中扮演重要角色,具有创新驱动、无供应商锁定、社区协作和透明度等优势。 3. **AI/ML在企业中的应用挑战**:企业面临模型部署、基础设施、流程和文化等方面的挑战,其中模型部署是最大挑战。 4. **AI应用开发**:开发生成式AI应用涉及寻找LLM、尝试提示、实验数据、连接数据源、模型服务、异常处理、有限微调、RAG、端点评估、基准测试和监控等步骤。 5. **模型选择与定制**:根据用例选择合适的模型,如Instruct、Base、Embed等,并通过微调(Fine-tuning)和检索增强生成(RAG)进行定制。 6. **AI平台**:Red Hat提供集成的AI平台,支持模型开发、训练、部署、监控和存储,帮助企业和开发者实现AI价值。
"开源AI,企业如何用?" "AI模型定制,开源如何助力?" "AI应用开发,开源平台哪家强?"
客服
商务合作
小程序
服务号
折叠