- @geoand review 1

core/runtime/src/main/java/io/quarkiverse/langchain4j/data/AiStatsMessage.java

@@ -7,6 +7,13 @@
 import dev.langchain4j.internal.ValidationUtils;
 import dev.langchain4j.model.output.TokenUsage;
 
+/**
+ * This class is the equivalent of Langchain4j AiMessage.
+ * It contains the token usage from the response that produce this AiMessage.
+ * And add the possibility to update the text in case of text containing Tools Result variables.
+ * Needed for @{@link io.quarkiverse.langchain4j.runtime.aiservice.ToolsResultMemory}
+ * Example of usage in ExperimentalParallelToolsDelegate in Ollama model provider
+ */
 public class AiStatsMessage extends AiMessage {
     private String updatableText;
 
@@ -15,12 +22,12 @@ public class AiStatsMessage extends AiMessage {
     public AiStatsMessage(String text, TokenUsage tokenUsage) {
         super(text);
         this.updatableText = text;
-        this.tokenUsage = ValidationUtils.ensureNotNull(tokenUsage, "tokeUsage");
+        this.tokenUsage = ValidationUtils.ensureNotNull(tokenUsage, "tokenUsage");
     }
 
     AiStatsMessage(List<ToolExecutionRequest> toolExecutionRequests, TokenUsage tokenUsage) {
         super(toolExecutionRequests);
-        this.tokenUsage = ValidationUtils.ensureNotNull(tokenUsage, "tokeUsage");
+        this.tokenUsage = ValidationUtils.ensureNotNull(tokenUsage, "tokenUsage");
     }
 
     AiStatsMessage(String text, List<ToolExecutionRequest> toolExecutionRequests, TokenUsage tokenUsage) {

core/runtime/src/main/java/io/quarkiverse/langchain4j/runtime/aiservice/AiServiceMethodImplementationSupport.java

@@ -27,7 +27,11 @@
 import dev.langchain4j.agent.tool.ToolExecutionRequest;
 import dev.langchain4j.agent.tool.ToolExecutor;
 import dev.langchain4j.agent.tool.ToolSpecification;
-import dev.langchain4j.data.message.*;
+import dev.langchain4j.data.message.AiMessage;
+import dev.langchain4j.data.message.ChatMessage;
+import dev.langchain4j.data.message.SystemMessage;
+import dev.langchain4j.data.message.ToolExecutionResultMessage;
+import dev.langchain4j.data.message.UserMessage;
 import dev.langchain4j.memory.ChatMemory;
 import dev.langchain4j.model.input.Prompt;
 import dev.langchain4j.model.input.PromptTemplate;
@@ -200,6 +204,7 @@ public void accept(Response<AiMessage> message) {
             AiMessage aiMessage = response.content();
 
             if (!aiMessage.hasToolExecutionRequests()) {
+                // If there is no tool Execution request we add the Ai Message directly in the chatMEmory
                 if (context.hasChatMemory()) {
                     context.chatMemory(memoryId).add(aiMessage);
                 }
@@ -228,6 +233,8 @@ public void accept(Response<AiMessage> message) {
                 }
                 tmpToolExecutionResultMessages.add(toolExecutionResultMessage);
             }
+            // In case of tool Execution request we need to update the AiMessage with tools results
+            // before adding it into chatMemory
             aiMessage = toolsResultMemory.substituteAiMessage(aiMessage);
             if (context.hasChatMemory()) {
                 chatMemory.add(aiMessage);

core/runtime/src/main/java/io/quarkiverse/langchain4j/runtime/aiservice/ToolsResultMemory.java

@@ -10,6 +10,13 @@
 import dev.langchain4j.data.message.AiMessage;
 import io.quarkiverse.langchain4j.data.AiStatsMessage;
 
+/**
+ * This class associate the ToolExecutionResul to the associated variable
+ * identified by @{@link ToolExecutionRequest#id()}.
+ * It will use them after when tools inputs @{@link ToolExecutionRequest#arguments()} are based
+ * on previous result variable and when AiMessage text contains variables too.
+ * See usage in @{@link AiServiceMethodImplementationSupport}
+ */
 @Experimental
 public class ToolsResultMemory {
 

model-providers/ollama/deployment/src/main/java/io/quarkiverse/langchain4j/ollama/deployment/OllamaProcessor.java

@@ -23,7 +23,6 @@
 import io.quarkiverse.langchain4j.ollama.runtime.config.LangChain4jOllamaConfig;
 import io.quarkiverse.langchain4j.ollama.runtime.config.LangChain4jOllamaFixedRuntimeConfig;
 import io.quarkiverse.langchain4j.ollama.tool.ExperimentalParallelToolsDelegate;
-import io.quarkiverse.langchain4j.ollama.tool.ExperimentalSequentialToolsDelegate;
 import io.quarkiverse.langchain4j.ollama.tool.NoToolsDelegate;
 import io.quarkiverse.langchain4j.runtime.NamedConfigUtil;
 import io.quarkus.arc.deployment.SyntheticBeanBuildItem;
@@ -155,8 +154,7 @@ private void addQualifierIfNecessary(SyntheticBeanBuildItem.ExtendedBeanConfigur
 
     @BuildStep
     public void handleDelegates(BuildProducer<RuntimeInitializedClassBuildItem> runtimeInitializedClassProducer) {
-        Stream.of(ExperimentalSequentialToolsDelegate.class.getName(),
-                ExperimentalParallelToolsDelegate.class.getName(),
+        Stream.of(ExperimentalParallelToolsDelegate.class.getName(),
                 NoToolsDelegate.class.getName())
                 .map(RuntimeInitializedClassBuildItem::new)
                 .forEach(runtimeInitializedClassProducer::produce);

model-providers/ollama/runtime/src/main/java/io/quarkiverse/langchain4j/ollama/OllamaChatLanguageModel.java

@@ -12,7 +12,6 @@
 import dev.langchain4j.model.chat.ChatLanguageModel;
 import dev.langchain4j.model.output.Response;
 import io.quarkiverse.langchain4j.ollama.tool.ExperimentalParallelToolsDelegate;
-import io.quarkiverse.langchain4j.ollama.tool.ExperimentalSequentialToolsDelegate;
 import io.quarkiverse.langchain4j.ollama.tool.ExperimentalTools;
 import io.quarkiverse.langchain4j.ollama.tool.NoToolsDelegate;
 
@@ -36,7 +35,6 @@ public class OllamaChatLanguageModel implements ChatLanguageModel {
     private ChatLanguageModel getDelegate(ExperimentalTools toolsEnum) {
         return switch (toolsEnum) {
             case NONE -> new NoToolsDelegate(this.client, this.model, this.options, this.format);
-            case SEQUENTIAL -> new ExperimentalSequentialToolsDelegate(this.client, this.model, this.options);
             case PARALLEL -> new ExperimentalParallelToolsDelegate(this.client, this.model, this.options);
         };
     }

model-providers/ollama/runtime/src/main/java/io/quarkiverse/langchain4j/ollama/OllamaMessagesUtils.java

@@ -8,7 +8,13 @@
 import java.util.function.Predicate;
 import java.util.stream.Collectors;
 
-import dev.langchain4j.data.message.*;
+import dev.langchain4j.data.message.ChatMessage;
+import dev.langchain4j.data.message.ChatMessageType;
+import dev.langchain4j.data.message.Content;
+import dev.langchain4j.data.message.ContentType;
+import dev.langchain4j.data.message.ImageContent;
+import dev.langchain4j.data.message.TextContent;
+import dev.langchain4j.data.message.UserMessage;
 
 public class OllamaMessagesUtils {
 

model-providers/ollama/runtime/src/main/java/io/quarkiverse/langchain4j/ollama/runtime/config/LangChain4jOllamaConfig.java

@@ -74,7 +74,6 @@ interface OllamaConfig {
          * NONE: for no tools, works as before without handling tools.
          * PARALLEL: Tools we be used and simulated with one call to llm, that will answer with all tool request to execute
          * and the response using the result of the tool request.
-         * SEQUENTIAL: Tools will be call sequentially and llm will call next tool following tool request result.
          */
         @WithDefault("NONE")
         Optional<String> experimentalTools();

model-providers/ollama/runtime/src/main/java/io/quarkiverse/langchain4j/ollama/tool/ExperimentalMessagesUtils.java

@@ -17,6 +17,10 @@
 import io.quarkiverse.langchain4j.ollama.Message;
 import io.quarkiverse.langchain4j.ollama.Role;
 
+/**
+ * Same as @{@link io.quarkiverse.langchain4j.ollama.OllamaMessagesUtils} but with AiStatsMessage Support
+ * And allow to group Messages for Experimental Tool llm request.
+ */
 class ExperimentalMessagesUtils {
 
     final static Predicate<ChatMessage> isUserMessage = chatMessage -> chatMessage instanceof UserMessage;
@@ -42,7 +46,7 @@ static AiStatsMessage toAiStatsMessage(List<ChatMessage> messages) {
                 .findFirst().orElseThrow();
     }
 
-    static AiStatsMessage withoutRequests(AiStatsMessage aiStatsMessage) {
+    static AiStatsMessage withoutToolRequests(AiStatsMessage aiStatsMessage) {
         return new AiStatsMessage(aiStatsMessage.text(), aiStatsMessage.getTokenUsage());
     }
 

model-providers/ollama/runtime/src/main/java/io/quarkiverse/langchain4j/ollama/tool/ExperimentalParallelToolsDelegate.java

@@ -70,7 +70,16 @@ record ToolResponse(String name, Map<String, Object> inputs, String result_id) {
     public ExperimentalParallelToolsDelegate(OllamaClient client, String modelName, Options options) {
         this.client = client;
         this.modelName = modelName;
-        this.options = options;
+        this.options = Options.builder()
+                .topP(options.topP())
+                .topK(options.topK())
+                .numCtx(options.numCtx())
+                .numPredict(options.numPredict())
+                .seed(options.seed())
+                .stop(options.stop())
+                .repeatPenalty(options.repeatPenalty())
+                .temperature(0.0)
+                .build();
     }
 
     @Override
@@ -102,7 +111,7 @@ public Response<AiMessage> generate(List<ChatMessage> messages,
             if (aiStatsMessage.text() == null) {
                 throw new RuntimeException("Conclusion cannot be null or empty!");
             }
-            return Response.from(withoutRequests(aiStatsMessage), aiStatsMessage.getTokenUsage(), FinishReason.STOP);
+            return Response.from(withoutToolRequests(aiStatsMessage), aiStatsMessage.getTokenUsage(), FinishReason.STOP);
         }
 
         Message systemMessage = createSystemMessageWithTools(ollamaMessages, toolSpecifications);

model-providers/ollama/runtime/src/main/java/io/quarkiverse/langchain4j/ollama/tool/ExperimentalTools.java

@@ -2,6 +2,5 @@
 
 public enum ExperimentalTools {
     NONE,
-    SEQUENTIAL,
     PARALLEL
 }