Added LLaMA.c++ as local option

devoxx · Aug 2, 2024 · 6dd8e73 · 6dd8e73
1 parent c0243f6
commit 6dd8e73
Show file tree

Hide file tree

Showing 10 changed files with 27 additions and 63 deletions.
diff --git a/src/main/java/com/devoxx/genie/chatmodel/ChatModelFactoryProvider.java b/src/main/java/com/devoxx/genie/chatmodel/ChatModelFactoryProvider.java
@@ -3,7 +3,7 @@
 import com.devoxx.genie.chatmodel.anthropic.AnthropicChatModelFactory;
 import com.devoxx.genie.chatmodel.deepinfra.DeepInfraChatModelFactory;
 import com.devoxx.genie.chatmodel.exo.ExoChatModelFactory;
-import com.devoxx.genie.chatmodel.gemini.GeminiChatModelFactory;
+import com.devoxx.genie.chatmodel.google.GoogleChatModelFactory;
 import com.devoxx.genie.chatmodel.groq.GroqChatModelFactory;
 import com.devoxx.genie.chatmodel.mistral.MistralChatModelFactory;
 import com.devoxx.genie.chatmodel.ollama.OllamaChatModelFactory;
@@ -28,7 +28,7 @@ public class ChatModelFactoryProvider {
         ModelProvider.Mistral.getName(), MistralChatModelFactory::new,
         ModelProvider.Groq.getName(), GroqChatModelFactory::new,
         ModelProvider.DeepInfra.getName(), DeepInfraChatModelFactory::new,
-        ModelProvider.Google.getName(), GeminiChatModelFactory::new
+        ModelProvider.Google.getName(), GoogleChatModelFactory::new
     );
 
     /**

diff --git a/src/main/java/com/devoxx/genie/chatmodel/ChatModelProvider.java b/src/main/java/com/devoxx/genie/chatmodel/ChatModelProvider.java
@@ -2,7 +2,7 @@
 
 import com.devoxx.genie.chatmodel.anthropic.AnthropicChatModelFactory;
 import com.devoxx.genie.chatmodel.exo.ExoChatModelFactory;
-import com.devoxx.genie.chatmodel.gemini.GeminiChatModelFactory;
+import com.devoxx.genie.chatmodel.google.GoogleChatModelFactory;
 import com.devoxx.genie.chatmodel.gpt4all.GPT4AllChatModelFactory;
 import com.devoxx.genie.chatmodel.groq.GroqChatModelFactory;
 import com.devoxx.genie.chatmodel.llama.LlamaChatModelFactory;
@@ -19,9 +19,7 @@
 import dev.langchain4j.model.chat.ChatLanguageModel;
 import dev.langchain4j.model.chat.StreamingChatLanguageModel;
 import lombok.Setter;
-import org.jetbrains.annotations.Contract;
 import org.jetbrains.annotations.NotNull;
-import org.jetbrains.annotations.Nullable;
 
 import java.util.HashMap;
 import java.util.Map;
@@ -43,9 +41,10 @@ public ChatModelProvider() {
         factories.put(ModelProvider.Mistral, new MistralChatModelFactory());
         factories.put(ModelProvider.Anthropic, new AnthropicChatModelFactory());
         factories.put(ModelProvider.Groq, new GroqChatModelFactory());
-        factories.put(ModelProvider.Google, new GeminiChatModelFactory());
+        factories.put(ModelProvider.Google, new GoogleChatModelFactory());
         factories.put(ModelProvider.Exo, new ExoChatModelFactory());
         factories.put(ModelProvider.LLaMA, new LlamaChatModelFactory());
+
         // TODO Currently broken by latest Jan! version
         // factories.put(ModelProvider.Jan, new JanChatModelFactory());
     }
@@ -114,29 +113,6 @@ private void setLocalBaseUrl(@NotNull LanguageModel languageModel,
         }
     }
 
-//    // TODO: This method is duplicated in multiple places. Consider moving it to a common utility class.
-//    private String getModelName(@Nullable LanguageModel languageModel) {
-//        if (languageModel == null) {
-//            return getDefaultModelName(null);
-//        }
-//
-//        return Optional.ofNullable(languageModel.getModelName())
-//            .orElseGet(() -> getDefaultModelName(languageModel.getProvider()));
-//    }
-//
-//    @Contract(pure = true)
-//    private @NotNull String getDefaultModelName(@Nullable ModelProvider provider) {
-//        if (provider == null) {
-//            return TEST_MODEL;
-//        }
-//        return switch (provider) {
-//            case LMStudio -> "LMStudio";
-//            case GPT4All -> "GPT4All";
-//            case LLaMA -> "LLaMA";
-//            default -> "DefaultModel";
-//        };
-//    }
-
     private static void setMaxOutputTokens(@NotNull DevoxxGenieStateService settingsState,
                                            @NotNull ChatModel chatModel) {
         Integer maxOutputTokens = settingsState.getMaxOutputTokens();

diff --git a/...tmodel/gemini/GeminiChatModelFactory.java → ...tmodel/google/GoogleChatModelFactory.java b/...tmodel/gemini/GeminiChatModelFactory.java → ...tmodel/google/GoogleChatModelFactory.java
@@ -1,4 +1,4 @@
-package com.devoxx.genie.chatmodel.gemini;
+package com.devoxx.genie.chatmodel.google;
 
 import com.devoxx.genie.chatmodel.ChatModelFactory;
 import com.devoxx.genie.model.ChatModel;
@@ -12,7 +12,7 @@
 import java.time.Duration;
 import java.util.List;
 
-public class GeminiChatModelFactory implements ChatModelFactory {
+public class GoogleChatModelFactory implements ChatModelFactory {
 
     @Override
     public ChatLanguageModel createChatModel(@NotNull ChatModel chatModel) {

diff --git a/src/main/java/com/devoxx/genie/model/GenericOpenAIProvider.java b/src/main/java/com/devoxx/genie/model/GenericOpenAIProvider.java
@@ -0,0 +1,14 @@
+package com.devoxx.genie.model;
+
+import lombok.Data;
+
+@Data
+public class GenericOpenAIProvider {
+    private String name;
+    private String baseUrl;
+    private String modelName;
+    private String apiKey;
+    private Double inputCost;
+    private Double outputCost;
+    private Integer contextWindow;
+}
diff --git a/src/main/java/com/devoxx/genie/model/enumarations/ModelProvider.java b/src/main/java/com/devoxx/genie/model/enumarations/ModelProvider.java
@@ -16,7 +16,7 @@ public enum ModelProvider {
     DeepInfra("DeepInfra"),
     Google("Google"),
     Exo("Exo (Experimental)"),
-    LLaMA("LLaMA.c++"),;
+    LLaMA("LLaMA.c++");
 
     private final String name;
 

diff --git a/src/main/java/com/devoxx/genie/service/LLMModelRegistryService.java b/src/main/java/com/devoxx/genie/service/LLMModelRegistryService.java
@@ -446,15 +446,6 @@ public void setModels(List<LanguageModel> models) {
         this.models = new ArrayList<>(models);
     }
 
-    public void addModel(LanguageModel model) {
-        models.add(model);
-    }
-
-    public void updateModel(LanguageModel updateModel) {
-        models.removeIf(model -> model.getModelName().equals(updateModel.getModelName()));
-        models.add(updateModel);
-    }
-
     @Override
     public LLMModelRegistryService getState() {
         return this;

diff --git a/src/main/java/com/devoxx/genie/ui/settings/DevoxxGenieStateService.java b/src/main/java/com/devoxx/genie/ui/settings/DevoxxGenieStateService.java
@@ -137,15 +137,6 @@ public void setModelCost(ModelProvider provider,
         }
     }
 
-//    public double getModelInputCost(ModelProvider provider, String modelName) {
-//        if (DefaultLLMSettingsUtil.isApiBasedProvider(provider)) {
-//            String key = provider.getName() + ":" + modelName;
-//            return modelInputCosts.getOrDefault(key,
-//                DefaultLLMSettingsUtil.DEFAULT_INPUT_COSTS.getOrDefault(new DefaultLLMSettingsUtil.CostKey(provider, modelName), 0.0));
-//        }
-//        return 0.0;
-//    }
-
     public double getModelInputCost(@NotNull ModelProvider provider, String modelName) {
         String key = provider.getName() + ":" + modelName;
         double cost = modelInputCosts.getOrDefault(key, 0.0);
@@ -165,15 +156,6 @@ public double getModelInputCost(@NotNull ModelProvider provider, String modelNam
         return cost;
     }
 
-    public double getModelOutputCost(ModelProvider provider, String modelName) {
-        if (DefaultLLMSettingsUtil.isApiBasedProvider(provider)) {
-            String key = provider.getName() + ":" + modelName;
-            return modelOutputCosts.getOrDefault(key,
-                DefaultLLMSettingsUtil.DEFAULT_OUTPUT_COSTS.getOrDefault(new DefaultLLMSettingsUtil.CostKey(provider, modelName), 0.0));
-        }
-        return 0.0;
-    }
-
     private void initializeDefaultCostsIfEmpty() {
         if (modelInputCosts.isEmpty()) {
             for (Map.Entry<DefaultLLMSettingsUtil.CostKey, Double> entry : DefaultLLMSettingsUtil.DEFAULT_INPUT_COSTS.entrySet()) {

diff --git a/src/main/resources/META-INF/plugin.xml b/src/main/resources/META-INF/plugin.xml
@@ -38,6 +38,7 @@
         <h2>v0.2.10</h2>
         <UL>
             <LI>Fix #184 - Input panel has bigger min/preferred height size</LI>
+            <LI>Feat #186 - Support for local LLaMA.c++ http server</LI>
         </UL>
         <h2>v0.2.9</h2>
         <UL>

diff --git a/src/main/resources/application.properties b/src/main/resources/application.properties
@@ -1,2 +1,2 @@
-#Thu Aug 01 20:57:17 CEST 2024
+#Fri Aug 02 17:59:31 CEST 2024
 version=0.2.10
diff --git a/...el/gemini/GeminiChatModelFactoryTest.java → ...el/google/GeminiChatModelFactoryTest.java b/...el/gemini/GeminiChatModelFactoryTest.java → ...el/google/GeminiChatModelFactoryTest.java
@@ -1,4 +1,4 @@
-package com.devoxx.genie.chatmodel.gemini;
+package com.devoxx.genie.chatmodel.google;
 
 import com.devoxx.genie.chatmodel.AbstractLightPlatformTestCase;
 import com.devoxx.genie.model.ChatModel;
@@ -37,7 +37,7 @@ public void setUp() throws Exception {
     @Test
     public void createChatModel() {
         // Instance of the class containing the method to be tested
-        var factory = new GeminiChatModelFactory();
+        var factory = new GoogleChatModelFactory();
 
         // Create a dummy ChatModel
         ChatModel chatModel = new ChatModel();
@@ -54,7 +54,7 @@ public void createChatModel() {
 
     @Test
     public void testModelNames() {
-        GeminiChatModelFactory factory = new GeminiChatModelFactory();
+        GoogleChatModelFactory factory = new GoogleChatModelFactory();
         Assertions.assertThat(factory.getModels()).isNotEmpty();
 
         List<LanguageModel> modelNames = factory.getModels();