Update java-llama to 2.0

Author: AutonomicPerfectionist

pom.xml

@@ -13,6 +13,9 @@
     # fast build
     mvn -DskipTests package -o 
 
+    # execute
+    mvn exec:java -Dexec.mainClass=org.myrobotlab.service.Runtime -Dexec.args="-s webgui WebGui intro Intro python Python"
+
     # specific test
     mvn test -Dtest="org.myrobotlab.service.WebGuiTest#postTest"  
 
@@ -80,7 +83,7 @@
       <!-- force overriding property at command line, use ${maven.build.timestamp}-->
       <timestamp>${maven.build.timestamp}</timestamp>
       <maven.build.timestamp.format>yyyyMMddHHmm</maven.build.timestamp.format>      
-      <version>${version}</version>
+      <version>${version}</version>   
       <GitBranch>${git.branch}</GitBranch>
       <username>${NODE_NAME}</username>      
       <platform>${NODE_LABELS}</platform>      
@@ -618,7 +621,7 @@
   <dependency>
     <groupId>de.kherud</groupId>
     <artifactId>llama</artifactId>
-    <version>1.1.4</version>
+    <version>2.2.1</version>
     <scope>provided</scope>
   </dependency>
 <!-- Llama end -->
@@ -1027,6 +1030,12 @@
     <artifactId>jovr</artifactId>
     <version>1.8.0.0</version>
     <scope>provided</scope>
+    <exclusions>
+      <exclusion>
+        <groupId>net.java.dev.jna</groupId>
+        <artifactId>jna</artifactId>
+      </exclusion>
+    </exclusions>
   </dependency>
   <dependency>
     <groupId>slick-util</groupId>

src/main/java/org/myrobotlab/service/Llama.java

@@ -1,7 +1,8 @@
 package org.myrobotlab.service;
 
+import de.kherud.llama.InferenceParameters;
 import de.kherud.llama.LlamaModel;
-import de.kherud.llama.Parameters;
+import de.kherud.llama.ModelParameters;
 import org.myrobotlab.framework.Platform;
 import org.myrobotlab.framework.Service;
 import org.myrobotlab.logging.Level;
@@ -36,25 +37,29 @@ public Llama(String reservedKey, String inId) {
     }
 
     public void loadModel(String modelPath) {
-        Parameters params = new Parameters.Builder()
+        ModelParameters params = new ModelParameters.Builder()
                 .setNGpuLayers(0)
                 .setNThreads(Platform.getLocalInstance().getNumPhysicalProcessors())
-                .setTemperature(0.7f)
-                .setPenalizeNl(true)
-                .setMirostat(Parameters.MiroStat.V2)
-                .setAntiPrompt(new String[]{config.userPrompt})
                 .build();
         model = new LlamaModel(modelPath, params);
     }
 
     public Response getResponse(String text) {
+        InferenceParameters inferenceParameters = new InferenceParameters.Builder()
+                .setTemperature(0.7f)
+                .setPenalizeNl(true)
+                .setMirostat(InferenceParameters.MiroStat.V2)
+                .setAntiPrompt(new String[]{config.userPrompt})
+                .build();
+
+
         if (model == null) {
             error("Model is not loaded.");
             return null;
         }
 
         String prompt = config.systemPrompt + config.systemMessage + "\n" + text + "\n";
-        String response = StreamSupport.stream(model.generate(prompt).spliterator(), false)
+        String response = StreamSupport.stream(model.generate(prompt, inferenceParameters).spliterator(), false)
                 .map(LlamaModel.Output::toString)
                 .reduce("", (a, b) -> a + b);
 
@@ -132,7 +137,7 @@ public Response publishResponse(Response response) {
     }
 
     public void reset() {
-        model.reset();
+        model.close();
 
     }
 

src/main/java/org/myrobotlab/service/meta/LlamaMeta.java

@@ -10,6 +10,6 @@ public LlamaMeta() {
                         "llama.cpp project. Can run most GGUF models."
         );
 
-        addDependency("de.kherud", "llama", "1.1.4");
+        addDependency("de.kherud", "llama", "2.2.1");
     }
 }