Flink-1.19.0源碼詳解-番外補充3-StreamGraph圖

1.StreamGraph圖:

? ? ? ??StreamGraph是Flink流處理作業的第一個計算調度流圖,它是從用戶編寫的 DataStream API程序轉換而來的邏輯圖。StreamGraph由StreamNode與StreamEdge組成,StreamNode為記錄數據處理的節點,StreamEdge為連接兩個StreamNode的邊。

StreamGraph圖解:

2.StreamNode節點:

? ? ? ??StreamNode是StreamGraph中的基本構建單元,代表了數據處理邏輯中的一個獨立操作或階段。每個StreamNode對應DataStream API中的一個轉換操作(Transformation)。

StreamNode圖解:

? ? ? ? StreamNode主要封裝了算子操作邏輯StreamOperator和其SimpleOperatorFactory,算子的入邊與出邊集合、還記錄SlotSharingGroup與ColocationGroup的等配置信息。

StreamNode完整源碼:

public class StreamNode {private final int id;//并行度private int parallelism;/*** Maximum parallelism for this stream node. The maximum parallelism is the upper limit for* dynamic scaling and the number of key groups used for partitioned state.*/private int maxParallelism;private ResourceSpec minResources = ResourceSpec.DEFAULT;private ResourceSpec preferredResources = ResourceSpec.DEFAULT;private final Map<ManagedMemoryUseCase, Integer> managedMemoryOperatorScopeUseCaseWeights =new HashMap<>();private final Set<ManagedMemoryUseCase> managedMemorySlotScopeUseCases = new HashSet<>();private long bufferTimeout;private final String operatorName;private String operatorDescription;//slotSharingGroup與coLocationGroup配置private @Nullable String slotSharingGroup;private @Nullable String coLocationGroup;private KeySelector<?, ?>[] statePartitioners = new KeySelector[0];private TypeSerializer<?> stateKeySerializer;//封裝算子的StreamOperatorFactoryprivate StreamOperatorFactory<?> operatorFactory;private TypeSerializer<?>[] typeSerializersIn = new TypeSerializer[0];private TypeSerializer<?> typeSerializerOut;//算子的入邊與出邊private List<StreamEdge> inEdges = new ArrayList<StreamEdge>();private List<StreamEdge> outEdges = new ArrayList<StreamEdge>();private final Class<? extends TaskInvokable> jobVertexClass;private InputFormat<?, ?> inputFormat;private OutputFormat<?> outputFormat;private String transformationUID;private String userHash;private final Map<Integer, StreamConfig.InputRequirement> inputRequirements = new HashMap<>();private @Nullable IntermediateDataSetID consumeClusterDatasetId;private boolean supportsConcurrentExecutionAttempts = true;private boolean parallelismConfigured = false;@VisibleForTestingpublic StreamNode(Integer id,@Nullable String slotSharingGroup,@Nullable String coLocationGroup,StreamOperator<?> operator,String operatorName,Class<? extends TaskInvokable> jobVertexClass) {this(id,slotSharingGroup,coLocationGroup,SimpleOperatorFactory.of(operator),operatorName,jobVertexClass);}public StreamNode(Integer id,@Nullable String slotSharingGroup,@Nullable String coLocationGroup,StreamOperatorFactory<?> operatorFactory,String operatorName,Class<? extends TaskInvokable> jobVertexClass) {this.id = id;this.operatorName = operatorName;this.operatorDescription = operatorName;this.operatorFactory = operatorFactory;this.jobVertexClass = jobVertexClass;this.slotSharingGroup = slotSharingGroup;this.coLocationGroup = coLocationGroup;}public void addInEdge(StreamEdge inEdge) {checkState(inEdges.stream().noneMatch(inEdge::equals),"Adding not unique edge = %s to existing inEdges = %s",inEdge,inEdges);if (inEdge.getTargetId() != getId()) {throw new IllegalArgumentException("Destination id doesn't match the StreamNode id");} else {inEdges.add(inEdge);}}public void addOutEdge(StreamEdge outEdge) {checkState(outEdges.stream().noneMatch(outEdge::equals),"Adding not unique edge = %s to existing outEdges = %s",outEdge,outEdges);if (outEdge.getSourceId() != getId()) {throw new IllegalArgumentException("Source id doesn't match the StreamNode id");} else {outEdges.add(outEdge);}}public List<StreamEdge> getOutEdges() {return outEdges;}public List<StreamEdge> getInEdges() {return inEdges;}public List<Integer> getOutEdgeIndices() {List<Integer> outEdgeIndices = new ArrayList<Integer>();for (StreamEdge edge : outEdges) {outEdgeIndices.add(edge.getTargetId());}return outEdgeIndices;}public List<Integer> getInEdgeIndices() {List<Integer> inEdgeIndices = new ArrayList<Integer>();for (StreamEdge edge : inEdges) {inEdgeIndices.add(edge.getSourceId());}return inEdgeIndices;}public int getId() {return id;}public int getParallelism() {return parallelism;}public void setParallelism(Integer parallelism) {setParallelism(parallelism, true);}void setParallelism(Integer parallelism, boolean parallelismConfigured) {this.parallelism = parallelism;this.parallelismConfigured =parallelismConfigured && parallelism != ExecutionConfig.PARALLELISM_DEFAULT;}/*** Get the maximum parallelism for this stream node.** @return Maximum parallelism*/int getMaxParallelism() {return maxParallelism;}/*** Set the maximum parallelism for this stream node.** @param maxParallelism Maximum parallelism to be set*/void setMaxParallelism(int maxParallelism) {this.maxParallelism = maxParallelism;}public ResourceSpec getMinResources() {return minResources;}public ResourceSpec getPreferredResources() {return preferredResources;}public void setResources(ResourceSpec minResources, ResourceSpec preferredResources) {this.minResources = minResources;this.preferredResources = preferredResources;}public void setManagedMemoryUseCaseWeights(Map<ManagedMemoryUseCase, Integer> operatorScopeUseCaseWeights,Set<ManagedMemoryUseCase> slotScopeUseCases) {managedMemoryOperatorScopeUseCaseWeights.putAll(operatorScopeUseCaseWeights);managedMemorySlotScopeUseCases.addAll(slotScopeUseCases);}public Map<ManagedMemoryUseCase, Integer> getManagedMemoryOperatorScopeUseCaseWeights() {return Collections.unmodifiableMap(managedMemoryOperatorScopeUseCaseWeights);}public Set<ManagedMemoryUseCase> getManagedMemorySlotScopeUseCases() {return Collections.unmodifiableSet(managedMemorySlotScopeUseCases);}public long getBufferTimeout() {return bufferTimeout;}public void setBufferTimeout(Long bufferTimeout) {this.bufferTimeout = bufferTimeout;}@VisibleForTestingpublic StreamOperator<?> getOperator() {return (StreamOperator<?>) ((SimpleOperatorFactory) operatorFactory).getOperator();}public StreamOperatorFactory<?> getOperatorFactory() {return operatorFactory;}public String getOperatorName() {return operatorName;}public String getOperatorDescription() {return operatorDescription;}public void setOperatorDescription(String operatorDescription) {this.operatorDescription = operatorDescription;}public void setSerializersIn(TypeSerializer<?>... typeSerializersIn) {checkArgument(typeSerializersIn.length > 0);// Unfortunately code above assumes type serializer can be null, while users of for example// getTypeSerializersIn would be confused by returning an array size of two with all// elements set to null...this.typeSerializersIn =Arrays.stream(typeSerializersIn).filter(typeSerializer -> typeSerializer != null).toArray(TypeSerializer<?>[]::new);}public TypeSerializer<?>[] getTypeSerializersIn() {return typeSerializersIn;}public TypeSerializer<?> getTypeSerializerOut() {return typeSerializerOut;}public void setSerializerOut(TypeSerializer<?> typeSerializerOut) {this.typeSerializerOut = typeSerializerOut;}public Class<? extends TaskInvokable> getJobVertexClass() {return jobVertexClass;}public InputFormat<?, ?> getInputFormat() {return inputFormat;}public void setInputFormat(InputFormat<?, ?> inputFormat) {this.inputFormat = inputFormat;}public OutputFormat<?> getOutputFormat() {return outputFormat;}public void setOutputFormat(OutputFormat<?> outputFormat) {this.outputFormat = outputFormat;}public void setSlotSharingGroup(@Nullable String slotSharingGroup) {this.slotSharingGroup = slotSharingGroup;}@Nullablepublic String getSlotSharingGroup() {return slotSharingGroup;}public void setCoLocationGroup(@Nullable String coLocationGroup) {this.coLocationGroup = coLocationGroup;}public @Nullable String getCoLocationGroup() {return coLocationGroup;}public boolean isSameSlotSharingGroup(StreamNode downstreamVertex) {return (slotSharingGroup == null && downstreamVertex.slotSharingGroup == null)|| (slotSharingGroup != null&& slotSharingGroup.equals(downstreamVertex.slotSharingGroup));}@Overridepublic String toString() {return operatorName + "-" + id;}public KeySelector<?, ?>[] getStatePartitioners() {return statePartitioners;}public void setStatePartitioners(KeySelector<?, ?>... statePartitioners) {checkArgument(statePartitioners.length > 0);this.statePartitioners = statePartitioners;}public TypeSerializer<?> getStateKeySerializer() {return stateKeySerializer;}public void setStateKeySerializer(TypeSerializer<?> stateKeySerializer) {this.stateKeySerializer = stateKeySerializer;}public String getTransformationUID() {return transformationUID;}void setTransformationUID(String transformationId) {this.transformationUID = transformationId;}public String getUserHash() {return userHash;}public void setUserHash(String userHash) {this.userHash = userHash;}public void addInputRequirement(int inputIndex, StreamConfig.InputRequirement inputRequirement) {inputRequirements.put(inputIndex, inputRequirement);}public Map<Integer, StreamConfig.InputRequirement> getInputRequirements() {return inputRequirements;}public Optional<OperatorCoordinator.Provider> getCoordinatorProvider(String operatorName, OperatorID operatorID) {if (operatorFactory instanceof CoordinatedOperatorFactory) {return Optional.of(((CoordinatedOperatorFactory) operatorFactory).getCoordinatorProvider(operatorName, operatorID));} else {return Optional.empty();}}boolean isParallelismConfigured() {return parallelismConfigured;}@Overridepublic boolean equals(Object o) {if (this == o) {return true;}if (o == null || getClass() != o.getClass()) {return false;}StreamNode that = (StreamNode) o;return id == that.id;}@Overridepublic int hashCode() {return id;}@Nullablepublic IntermediateDataSetID getConsumeClusterDatasetId() {return consumeClusterDatasetId;}public void setConsumeClusterDatasetId(@Nullable IntermediateDataSetID consumeClusterDatasetId) {this.consumeClusterDatasetId = consumeClusterDatasetId;}public boolean isSupportsConcurrentExecutionAttempts() {return supportsConcurrentExecutionAttempts;}public void setSupportsConcurrentExecutionAttempts(boolean supportsConcurrentExecutionAttempts) {this.supportsConcurrentExecutionAttempts = supportsConcurrentExecutionAttempts;}
}

3.StreamEdge邊:

? ? ? ? StreamEdge是StreamGraph中連接各個StreamNode的邊,它定義了數據在StreamNode節點之間的流動方式和分區策略。

StreamEdge圖解:

? ? ? ? StreamEdge主要封裝了StreamEdge連接的前后StreamNode的id,并記錄確定分區選擇的Partitioner。

StreamEdge源碼:

public class StreamEdge implements Serializable {private static final long serialVersionUID = 1L;private static final long ALWAYS_FLUSH_BUFFER_TIMEOUT = 0L;private final String edgeId;//封裝前后StreamNode節點的idprivate final int sourceId;private final int targetId;/*** Note that this field doesn't have to be unique among all {@link StreamEdge}s. It's enough if* this field ensures that all logical instances of {@link StreamEdge} are unique, and {@link* #hashCode()} are different and {@link #equals(Object)} returns false, for every possible pair* of {@link StreamEdge}. Especially among two different {@link StreamEdge}s that are connecting* the same pair of nodes.*/private final int uniqueId;/** The type number of the input for co-tasks. */private final int typeNumber;/** The side-output tag (if any) of this {@link StreamEdge}. */private final OutputTag outputTag;//分區器 StreamPartitioner/** The {@link StreamPartitioner} on this {@link StreamEdge}. */private StreamPartitioner<?> outputPartitioner;/** The name of the operator in the source vertex. */private final String sourceOperatorName;/** The name of the operator in the target vertex. */private final String targetOperatorName;private final StreamExchangeMode exchangeMode;private long bufferTimeout;private boolean supportsUnalignedCheckpoints = true;private final IntermediateDataSetID intermediateDatasetIdToProduce;public StreamEdge(StreamNode sourceVertex,StreamNode targetVertex,int typeNumber,StreamPartitioner<?> outputPartitioner,OutputTag outputTag) {this(sourceVertex,targetVertex,typeNumber,ALWAYS_FLUSH_BUFFER_TIMEOUT,outputPartitioner,outputTag,StreamExchangeMode.UNDEFINED,0,null);}public StreamEdge(StreamNode sourceVertex,StreamNode targetVertex,int typeNumber,StreamPartitioner<?> outputPartitioner,OutputTag outputTag,StreamExchangeMode exchangeMode,int uniqueId,IntermediateDataSetID intermediateDatasetId) {this(sourceVertex,targetVertex,typeNumber,sourceVertex.getBufferTimeout(),outputPartitioner,outputTag,exchangeMode,uniqueId,intermediateDatasetId);}public StreamEdge(StreamNode sourceVertex,StreamNode targetVertex,int typeNumber,long bufferTimeout,StreamPartitioner<?> outputPartitioner,OutputTag outputTag,StreamExchangeMode exchangeMode,int uniqueId,IntermediateDataSetID intermediateDatasetId) {this.sourceId = sourceVertex.getId();this.targetId = targetVertex.getId();this.uniqueId = uniqueId;this.typeNumber = typeNumber;this.bufferTimeout = bufferTimeout;this.outputPartitioner = outputPartitioner;this.outputTag = outputTag;this.sourceOperatorName = sourceVertex.getOperatorName();this.targetOperatorName = targetVertex.getOperatorName();this.exchangeMode = checkNotNull(exchangeMode);this.intermediateDatasetIdToProduce = intermediateDatasetId;this.edgeId =sourceVertex+ "_"+ targetVertex+ "_"+ typeNumber+ "_"+ outputPartitioner+ "_"+ uniqueId;}public int getSourceId() {return sourceId;}public int getTargetId() {return targetId;}public int getTypeNumber() {return typeNumber;}public OutputTag getOutputTag() {return this.outputTag;}public StreamPartitioner<?> getPartitioner() {return outputPartitioner;}public StreamExchangeMode getExchangeMode() {return exchangeMode;}public void setPartitioner(StreamPartitioner<?> partitioner) {this.outputPartitioner = partitioner;}public void setBufferTimeout(long bufferTimeout) {checkArgument(bufferTimeout >= -1);this.bufferTimeout = bufferTimeout;}public long getBufferTimeout() {return bufferTimeout;}public void setSupportsUnalignedCheckpoints(boolean supportsUnalignedCheckpoints) {this.supportsUnalignedCheckpoints = supportsUnalignedCheckpoints;}public boolean supportsUnalignedCheckpoints() {return supportsUnalignedCheckpoints;}@Overridepublic int hashCode() {return Objects.hash(edgeId, outputTag);}@Overridepublic boolean equals(Object o) {if (this == o) {return true;}if (o == null || getClass() != o.getClass()) {return false;}StreamEdge that = (StreamEdge) o;return Objects.equals(edgeId, that.edgeId) && Objects.equals(outputTag, that.outputTag);}@Overridepublic String toString() {return "("+ (sourceOperatorName + "-" + sourceId)+ " -> "+ (targetOperatorName + "-" + targetId)+ ", typeNumber="+ typeNumber+ ", outputPartitioner="+ outputPartitioner+ ", exchangeMode="+ exchangeMode+ ", bufferTimeout="+ bufferTimeout+ ", outputTag="+ outputTag+ ", uniqueId="+ uniqueId+ ')';}public IntermediateDataSetID getIntermediateDatasetIdToProduce() {return intermediateDatasetIdToProduce;}
}

? ? ? ? 本文是對StreamGraph的解釋與補充,完整StreamGraph創建源碼見《Flink-1.19.0源碼詳解4-StreamGraph生成》。

本文來自互聯網用戶投稿,該文觀點僅代表作者本人,不代表本站立場。本站僅提供信息存儲空間服務,不擁有所有權,不承擔相關法律責任。
如若轉載,請注明出處:http://www.pswp.cn/web/87249.shtml
繁體地址,請注明出處:http://hk.pswp.cn/web/87249.shtml
英文地址,請注明出處:http://en.pswp.cn/web/87249.shtml

如若內容造成侵權/違法違規/事實不符,請聯系多彩編程網進行投訴反饋email:809451989@qq.com,一經查實,立即刪除!

相關文章

linux系統---Nginx反向代理與緩存功能

目錄 正向代理和反向代理 正向代理的作用 反向代理可實現的功能 反向代理客戶端ip透傳 1.初始訪問192.168.235.139 結果 2.編輯代理服務器的配置文件 3、重載nginx服務 4、訪問代理服務器 實現反向代理負載均衡 1.先啟用已用另一臺服務端 2.使用192.168.235.140 …

U+平臺配置免密登錄、安裝Hadoop配置集群、Spark配置

文章目錄 1、免密登錄2、安裝hadoop3、Spark配置 具體詳細報告見資源部分&#xff0c;全部實驗內容已經上傳&#xff0c;如有需要請自行下載。 1、免密登錄 使用的配置命令&#xff1a; cd ~/.ssh/ssh-keygen -t rsaEnter鍵回車y回車回車出現如上所示 cat ./id_rsa.pub >…

GitHub vs GitLab 全面對比報告(2025版)

從技術架構到金融估值&#xff0c;深度解析兩大代碼托管平臺的差異化競爭策略 一、技術架構對比 維度GitHub (Microsoft旗下)GitLab (獨立上市公司)關鍵差異核心架構- 分布式Git倉庫 Issues/Projects- 全棧DevSecOps平臺GitLab集成CI/CD、安全、監控部署模式- SaaS為主 - Git…

Python 數據分析與可視化 Day 14 - 建模復盤 + 多模型評估對比(邏輯回歸 vs 決策樹)

? 今日目標 回顧整個本周數據分析 & 建模流程學會訓練第二種模型&#xff1a;決策樹&#xff08;Decision Tree&#xff09;掌握多模型對比評估的方法與實踐輸出綜合對比報告&#xff1a;準確率、精確率、召回率、F1 等指標為后續模型調優與擴展打下基礎 &#x1fa9c; 一…

本周大模型新動向:KV緩存混合精度量化、個體時空行為生成、個性化問答

點擊藍字 關注我們 AI TIME歡迎每一位AI愛好者的加入&#xff01; 01 KVmix: Gradient-Based Layer Importance-Aware Mixed-Precision Quantization for KV Cache 大型語言模型&#xff08;LLMs&#xff09;在推理過程中&#xff0c;鍵值&#xff08;KV&#xff09;緩存的高內…

在 Spring Boot 中使用 WebMvcConfigurer

WebMvcConfigurer 是 Spring MVC 提供的一個擴展接口&#xff0c;用于配置 Spring MVC 的各種功能。在 Spring Boot 應用中&#xff0c;通過實現 WebMvcConfigurer 接口&#xff0c;可以定制和擴展默認的 Spring MVC 配置。以下是對 WebMvcConfigurer 的詳細解析及其常見用法。…

w-筆記:uni-app的H5平臺和非H5平臺的拍照識別功能:

uni-app的H5平臺和非H5平臺的拍照識別功能&#xff1a; <template><view class"humanVehicleBinding"><view v-if"warn" class"shadow"></view><view class"header"><uni-nav-bar left-icon"l…

TCP 半連接隊列和全連接隊列(結合 Linux 2.6.32 內核源碼分析)

文章目錄 一、什么是 TCP 半連接隊列和全連接隊列二、TCP 全連接隊列1、如何查看進程的 TCP 全連接隊列大小&#xff1f;注意 2、TCP 全連接隊列溢出問題注意 3、TCP 全連接隊列最大長度 三、TCP 半連接隊列1、TCP 半連接隊列溢出問題2、TCP 半連接隊列最大長度3、引申問題 一、…

linux下fabric環境搭建

參考教程&#xff1a; https://devpress.csdn.net/cloudnative/66d58e702045de334a569db3.html?dp_tokeneyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJpZCI6MjA2MzY4NywiZXhwIjoxNzQwMzY4MDc0LCJpYXQiOjE3Mzk3NjMyNzQsInVzZXJuYW1lIjoiaHVhbmd0dXBpIn0.oh8e4F6Sw_A4SV2ODQ5W0pYK0…

Redis Pipeline介紹:提高操作Redis數據庫的執行效率

Redis Pipeline是一種用于提高Redis執行效率的技術&#xff0c;通過減少客戶端與服務器之間的通信開銷&#xff0c;顯著提升批量操作的性能。本文將詳細介紹Redis Pipeline的概念、使用場景、實現方式及其優勢。 一、Redis Pipeline的概念 Redis Pipeline是一種批處理機制&am…

linux長時間鎖屏無法喚醒

是的&#xff0c;您這么理解很直接&#xff0c;抓住了要點。 簡單來說&#xff0c;就是這樣&#xff1a; 電腦睡覺有兩種方式&#xff1a; 打個盹&#xff08;掛起/Suspend&#xff09;&#xff1a; 把工作狀態保存在內存里。這個一般和 Swap 分區沒關系。睡死過去&#xff…

STM32F103_Bootloader程序開發11 - 實現 App 安全跳轉至 Bootloader

導言 想象一下&#xff0c;我們的單片機 App 正在穩定地運行著&#xff0c;突然我們想給它升級一下&#xff0c;添加個新功能。我們該如何安全地通知它&#xff1a;“嘿&#xff0c;準備好接收新固件了” ? 這就需要 App 和 Bootloader 之間建立一個可靠的"秘密握手"…

Explain解釋

參考官方文檔&#xff1a;https://dev.mysql.com/doc/refman/5.7/en/explain-output.html explain關鍵字可以分析你的查詢語句的結構和性能。 explain select查詢&#xff0c; 執行會返回執行計劃的信息。 注意&#xff1a;如果from中有子查詢&#xff0c;仍然會執行該子查詢…

選擇 PDF 轉 HTML 轉換器的 5 個關鍵特性

市面上有很多 PDF 轉 HTML 的轉換器&#xff0c;每一款產品都有不同的功能組合。要理清并理解每個功能可能會讓人感到困惑。那么&#xff0c;真正重要的是什么呢&#xff1f; 這篇文章將介紹我們認為在選擇最佳 PDF 轉 HTML 轉換器時最重要的 5 個關鍵特性&#xff1a; 1. 轉換…

使用堡塔在服務器上部署寶塔面板-linux版

使用堡塔在服務器上部署寶塔面板-linux版 使用堡塔多機管理登錄服務器 進入寶塔官網&#xff0c;獲取安裝腳本 wget -O install_panel.sh https://download.bt.cn/install/install_panel.sh && sudo bash install_panel.sh ed8484bec3. 在堡塔多機管理中&#xff0c;…

【Unity高級】Unity多界面游戲場景管理方案詳解

引言&#xff1a;游戲界面管理的挑戰 在Unity游戲開發中&#xff0c;尤其是包含多個功能界面&#xff08;如主菜單、關卡選擇、游戲頁面、設置和商城&#xff09;的游戲&#xff0c;如何高效管理場景與界面是架構設計的核心挑戰。本文將深入探討三種主流實現方案&#xff1a;單…

WINDOWS最快布署WEB服務器:apache2

安裝JDK下載 https://tomcat.apache.org/ Index of /dist/tomcat/tomcat-9 安裝測試 http://localhost:8080/ 替換自己的文件 把自己的文件復制到&#xff1a; C:\Program Files\Apache Software Foundation\Tomcat 9.0\webapps\ROOT

Microsoft Edge 打開無反應、打開后顯示兼容性問題、卸載重裝 解決方案。一鍵卸載Microsoft Edge 。

背景&#xff1a;網絡上的瀏覽器修復、重裝、恢復默認應用測試后無用&#xff0c;以下卸載重裝方案經實測可以正常使用Microsoft Edg。 卸載軟件在資源里&#xff0c;請自取。 一、卸載軟件&#xff1a;Remove-Edge_GUI.exe 雙擊卸載等待即可。 二、在微軟商店重新安裝Micro…

Spring Boot - 參數校驗:分組校驗、自定義注解、嵌套對象全解析

01 依賴配置 在構建高效的校驗體系前&#xff0c;需先完善項目依賴配置。 以下是優化后的依賴示例&#xff1a; <dependencies><!-- Web 依賴&#xff0c;提供 RESTful 接口支持 --><dependency><groupId>org.springframework.boot</groupId>…

深入淺出多模態》(十一)之多模態經典模型:Flamingo系列

&#x1f389;AI學習星球推薦&#xff1a; GoAI的學習社區 知識星球是一個致力于提供《機器學習 | 深度學習 | CV | NLP | 大模型 | 多模態 | AIGC 》各個最新AI方向綜述、論文等成體系的學習資料&#xff0c;配有全面而有深度的專欄內容&#xff0c;包括不限于 前沿論文解讀、…