使用私有 AI 增强您的 Ibexa DXP 项目.pdf

1、with Private AIAugumenting your Ibexa DXP ProjectMATEUSZ BIENIEKPrivate AIHugging Facehttps:/huggingface.coClick to edit Master title styleVery popular LLM by METAFree for(almost all)Commercial UseModel Sizes(1B,3B,11B,etc)Text GenerationIntended use case:commercial and reasearchuse in multiple lang

2、uagesStatic model tranined on an offline dataset(Sep,2024)Llama-3.2-3B https:/huggingface.co/meta-llama/Llama-3.2-3BOllama Pltaform for running LLMs locally Support for many popular models Easy to install and run Runs on everything”(Windows,Linux,MacOS)Support for multiple hardware(M-series CPUs fro

3、m apple,GPUs,etc)OpenSource and MIT license Pros and Cons of running AI locally https:/ and Ibexa DXPLTS Update(Opt-in feature)Integrating Features provided by AI services in yourprojectBackend UISupport for multiple connectionVery customizable and extendableActions,Action Types,Action Configuration

4、s,Handlersand more!Requirement:AI Actions!AI Framework for Ibexa DXP!Custom ConnectorSimple Custom AI Connector in 3 stepsGenerate Text with OllamaActionAction TypeHandlerLigths CameraACTIONLigths CameraACTIONLigths CameraACTION(Type)Ligths CameraHandler?Services!Demo 1!AI FunctionsAI FuntionsToolsP

5、rovide AI Agents with Dynamic DataAutomationsNot supported by all Modelstool_calls”responseAllow AI execute funtions in your code!New Tool”HandlerDemo 2!Semantic SearchSemantic SearchSemantic Search VS Keyword SearchVectorsEmbeddingsVector StoreSemantic search is a search engine technology that inte

6、rprets the meaning of words and phrases./Elastic.coVectorsEmbeddings Numerical representation of documents Captures semantics of the text Embedding of Question is highly similiarto the Answer Embedding models Generate vectors for text Embed