3 Gemma ..... 13 2.6.4 Llama3 ..... 13 3 零代码本地部署 AI 后端 ..... 13 3.1 大模型 Llama3 ..... 13 3.1.1 步骤 1：安装 Ollama ..... 13 3.1.2 步骤 2：安装 Llama ..... 14 3.1.3 使用 Llama3 ..... 15 3.2 大模型 phi-3 docker 下载 MaxKB 27 5.3 docker 配置 MaxKB 29 5.4 打开 MaxKB 网页 32 5.5 构建第一个私人知识库 34 5.6 MaxKB 配置本地 llama3 37 5.7 创建知识库应用 40 ## 1 AI 大模型基础 ### 1.1 AIGC AIGC 是指使用人工智能模型生成内容的技术。这些内容可以包括图像、音频、文本、视频、3D 模型等。具体来说，AIGC 闭源大模型包括 OpenAI 的 GPT 系列和 Google 的 BERT。这些模型因其高效的学习能力和强大的通用性而受到关注。 开源大模型以 Meta 的 Llama 系列，2024 年 4 月，Llama3 发布，包括 8B 和 70B 模型。 图 2，时间线主要根据技术论文的发布日期（例如提交至 arXiv 的日期）来确定大型语言模型（大小超过 10B）的发展历程。如果没有相应的论文，我们将模

including DeepSeek 67B (DeepSeek-AI, 2024) (our previous release), Qwen1.5 72B (Bai et al., 2023), LLaMA3 70B (AI@Meta, 2024), and Mixtral 8x22B (Mistral, 2024). We evaluate all these models with our internal Compared with LLaMA3 70B, DeepSeek-V2 is trained on fewer than a quarter of English tokens. Therefore, we acknowledge that DeepSeek-V2 still has a slight gap in basic English capabilities with LLaMA3 70B. However and activated parameters, DeepSeek-V2 still demonstrates comparable code and math capability with LLaMA3 70B. Also, as a bilingual language model, DeepSeek-V2 outperforms LLaMA3. 70B overwhelmingly on Chinese

2024年6月：Gemma2，性能跳了一大步 四个月后，2024年6月27日，Gemma2来了。首发两档：9B和27B，7月31日又追加了2B。 27B是Gemma2真正的亮点。在当时的开源模型里，27B这个尺寸段几乎没有竞品。Llama3的最小版本是 8B，往上就是70B，中间存在一个巨大的空白。Gemma227B刚好插进去。 性能提升也确实明显。在多个benchmark上，Gemma227B的表现接近甚至超过了一些70B级别的模型。

to="" model=""> is the location where the model is stored, for example, the weights for llama2 or llama3 models. ## 1.2.3 Option 2: Build from source 0. Install prerequisites (skip if you are already core42/jais-30b-chat-v3,etc.| |JambaForCausalLM|Jamba|ai21labs/Jamba-v0.1,etc.| |LlamaForCausalLM|Llama3.1,Llama3,Llama2,LLaMA,Yi|meta-llama/Meta-Llama-3.1-405B-Instruct,meta-llama/Meta-Llama-3.1-70B,meta-llama

to="" model=""> is the location where the model is stored, for example, the weights for llama2 or llama3 models. ## 1.2.3 Option 2: Build from source 0. Install prerequisites (skip if you are already core42/jais-30b-chat-v3,etc.| |JambaForCausalLM|Jamba|ai21labs/Jamba-v0.1,etc.| |LlamaForCausalLM|Llama3.1,Llama3,Llama2,LLaMA,Yi|meta-llama/Meta-Llama-3.1-405B-Instruct,meta-llama/Meta-Llama-3.1-70B,meta-llama

to="" model=""> is the location where the model is stored, for example, the weights for llama2 or llama3 models. ## 1.2.3 Option 2: Build from source 0. Install prerequisites (skip if you are already core42/jais-30b-chat-v3,etc.| |JambaForCausalLM|Jamba|ai21labs/Jamba-v0.1,etc.| |LlamaForCausalLM|Llama3.1,Llama3,Llama2,LLaMA,Yi|meta-llama/Meta-Llama-3.1-405B-Instruct,meta-llama/Meta-Llama-3.1-70B,meta-llama

to="" model=""> is the location where the model is stored, for example, the weights for llama2 or llama3 models. ## 1.2.3 Option 2: Build from source 0. Install prerequisites (skip if you are already

to="" model=""> is the location where the model is stored, for example, the weights for llama2 or llama3 models. ## 1.2.3 Option 2: Build from source 0. Install prerequisites (skip if you are already

to="" model=""> is the location where the model is stored, for example, the weights for llama2 or llama3 models. ## 1.2.3 Option 2: Build from source 0. Install prerequisites (skip if you are already

to="" model=""> is the location where the model is stored, for example, the weights for llama2 or llama3 models. ## 1.2.3 Option 2: Build from source 0. Install prerequisites (skip if you are already