GPU

.gitignore

@@ -0,0 +1,3 @@
+.idea/
+*.iml
+wireguard.txt

rag-service/pom.xml

@@ -64,9 +64,16 @@
 			<groupId>org.springframework.boot</groupId>
 			<artifactId>spring-boot-starter-web</artifactId>
 		</dependency>
+		<!-- Changed: spring-ai-starter-model-ollama -> spring-ai-starter-model-openai (Groq-compatible) -->
 		<dependency>
 			<groupId>org.springframework.ai</groupId>
-			<artifactId>spring-ai-starter-model-ollama</artifactId>
+			<artifactId>spring-ai-starter-model-openai</artifactId>
+		</dependency>
+		<!-- Local ONNX embedding model (Groq does not provide embedding API) -->
+		<!-- Default model: all-MiniLM-L6-v2 (384 dimensions) -->
+		<dependency>
+			<groupId>org.springframework.ai</groupId>
+			<artifactId>spring-ai-starter-model-transformers</artifactId>
 		</dependency>
 		<dependency>
 			<groupId>org.springframework.ai</groupId>

rag-service/src/main/java/com/balex/rag/RagApplication.java

@@ -14,7 +14,7 @@ import org.springframework.ai.chat.client.advisor.api.Advisor;
 import org.springframework.ai.chat.memory.ChatMemory;
 import org.springframework.ai.chat.model.ChatModel;
 import org.springframework.ai.chat.prompt.PromptTemplate;
-import org.springframework.ai.ollama.api.OllamaOptions;
+import org.springframework.ai.openai.OpenAiChatOptions;
 import org.springframework.ai.vectorstore.VectorStore;
 import org.springframework.beans.factory.annotation.Value;
 import org.springframework.boot.SpringApplication;
@@ -49,9 +49,10 @@ public class RagApplication {
 								.order(3).build(),
 						SimpleLoggerAdvisor.builder().order(4).build()
 				)
-				.defaultOptions(OllamaOptions.builder()
+				.defaultOptions(OpenAiChatOptions.builder()
 						.temperature(ragDefaults.temperature())
-						.repeatPenalty(ragDefaults.repeatPenalty())
+						.topP(ragDefaults.topP())
+						.frequencyPenalty(ragDefaults.repeatPenalty() - 1.0) // Ollama repeatPenalty 1.1 -> frequencyPenalty 0.1
 						.build())
 				.build();
 	}

rag-service/src/main/java/com/balex/rag/advisors/expansion/ExpansionQueryAdvisor.java

@@ -10,7 +10,7 @@ import org.springframework.ai.chat.client.advisor.api.AdvisorChain;
 import org.springframework.ai.chat.client.advisor.api.BaseAdvisor;
 import org.springframework.ai.chat.model.ChatModel;
 import org.springframework.ai.chat.prompt.PromptTemplate;
-import org.springframework.ai.ollama.api.OllamaOptions;
+import org.springframework.ai.openai.OpenAiChatOptions;
 
 import java.util.Map;
 
@@ -48,11 +48,10 @@ public class ExpansionQueryAdvisor implements BaseAdvisor {
 
     public static ExpansionQueryAdvisorBuilder builder(ChatModel chatModel, RagExpansionProperties props) {
         return new ExpansionQueryAdvisorBuilder().chatClient(ChatClient.builder(chatModel)
-                .defaultOptions(OllamaOptions.builder()
+                .defaultOptions(OpenAiChatOptions.builder()
                         .temperature(props.temperature())
-                        .topK(props.topK())
                         .topP(props.topP())
-                        .repeatPenalty(props.repeatPenalty())
+                        .frequencyPenalty(props.repeatPenalty() - 1.0) // Ollama repeatPenalty 1.0 -> frequencyPenalty 0.0
                         .build())
                 .build());
     }

rag-service/src/main/java/com/balex/rag/controller/ChatEntryController.java

@@ -36,16 +36,15 @@ public class ChatEntryController {
         log.trace(ApiLogMessage.NAME_OF_CURRENT_METHOD.getValue(), ApiUtils.getMethodName());
 
         boolean onlyContext = request.onlyContext() != null ? request.onlyContext() : ragDefaults.onlyContext();
-        int topK = request.topK() != null ? request.topK() : ragDefaults.topK();
         double topP = request.topP() != null ? request.topP() : ragDefaults.topP();
 
-        ChatEntry entry = chatEntryService.addUserEntry(chatId, request.content(), onlyContext, topK, topP);
+        ChatEntry entry = chatEntryService.addUserEntry(chatId, request.content(), onlyContext, topP);
 
         Chat chat = chatService.getChat(chatId);
         eventPublisher.publishQuerySent(
                 chat.getIdOwner().toString(),
                 chatId.toString(),
-                0);  // TODO: add tokensUsed when Ollama response provides it
+                0);  // TODO: add tokensUsed when usage info is available from Groq response
 
         return ResponseEntity.ok(entry);
     }

rag-service/src/main/java/com/balex/rag/model/dto/UserEntryRequest.java

@@ -3,6 +3,5 @@ package com.balex.rag.model.dto;
 public record UserEntryRequest(
         String content,
         Boolean onlyContext,
-        Integer topK,
         Double topP
 ) {}

rag-service/src/main/java/com/balex/rag/service/ChatEntryService.java

@@ -8,5 +8,5 @@ public interface ChatEntryService {
 
     List<ChatEntry> getEntriesByChatId(Long chatId);
 
-    ChatEntry addUserEntry(Long chatId, String content, boolean onlyContext, int topK, double topP);
+    ChatEntry addUserEntry(Long chatId, String content, boolean onlyContext, double topP);
 }

rag-service/src/main/java/com/balex/rag/service/impl/ChatEntryServiceImpl.java

@@ -11,7 +11,7 @@ import lombok.RequiredArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
 import org.springframework.ai.chat.client.ChatClient;
 import org.springframework.ai.chat.memory.ChatMemory;
-import org.springframework.ai.ollama.api.OllamaOptions;
+import org.springframework.ai.openai.OpenAiChatOptions;
 import org.springframework.stereotype.Service;
 import org.springframework.transaction.annotation.Transactional;
 
@@ -33,7 +33,7 @@ public class ChatEntryServiceImpl implements ChatEntryService {
 
     @Override
     @Transactional
-    public ChatEntry addUserEntry(Long chatId, String content, boolean onlyContext, int topK, double topP) {
+    public ChatEntry addUserEntry(Long chatId, String content, boolean onlyContext, double topP) {
         Chat chat = chatRepository.findById(chatId)
                 .orElseThrow(() -> new EntityNotFoundException("Chat not found with id: " + chatId));
 
@@ -63,8 +63,7 @@ public class ChatEntryServiceImpl implements ChatEntryService {
                 .system(systemPrompt)
                 .user(content)
                 .advisors(a -> a.param(ChatMemory.CONVERSATION_ID, String.valueOf(chatId)))
-                .options(OllamaOptions.builder()
+                .options(OpenAiChatOptions.builder()
-                        .topK(topK)
                         .topP(topP)
                         .build())
                 .call()

rag-service/src/main/resources/application.properties

@@ -1,24 +1,29 @@
-spring.application.name=rag
+spring.application.name=rag-service
-#spring.ai.ollama.base-url=http://localhost:11431
+
-spring.ai.ollama.base-url=http://localhost:11434
+# --- LLM Provider: Groq (OpenAI-compatible API) ---
-#spring.ai.ollama.chat.model=gemma3:4b-it-q4_K_M
+spring.ai.openai.base-url=${SPRING_AI_OPENAI_BASE_URL:https://api.groq.com/openai/v1}
-spring.ai.ollama.chat.model=llama3.1:8b-instruct-q4_K_M
+spring.ai.openai.api-key=${SPRING_AI_OPENAI_API_KEY:}
-jwt.secret=ywfI6dBznYmHbokihB/OBzZz6E0Fj+6PiqrM8dQ5c3t0HeYarblCbOGM8vQtOt472AtQ+MsCH7OVIKHOzjrPsQ==
+spring.ai.openai.chat.model=${SPRING_AI_OPENAI_CHAT_MODEL:llama-3.3-70b-versatile}
+
+# --- Embedding model: use local pgvector with a lightweight model ---
+# Groq does not provide an embedding endpoint, so we disable OpenAI embedding auto-config
+# and rely on the pgvector store's existing embedding setup
+spring.ai.openai.embedding.enabled=false
+
+jwt.secret=${JWT_SECRET:ywfI6dBznYmHbokihB/OBzZz6E0Fj+6PiqrM8dQ5c3t0HeYarblCbOGM8vQtOt472AtQ+MsCH7OVIKHOzjrPsQ==}
 jwt.expiration=103600000
-spring.datasource.url=jdbc:postgresql://localhost:5432/ragdb
+spring.datasource.url=${SPRING_DATASOURCE_URL:jdbc:postgresql://localhost:5432/ragdb}
-spring.datasource.username=postgres
+spring.datasource.username=${SPRING_DATASOURCE_USERNAME:postgres}
-spring.datasource.password=postgres
+spring.datasource.password=${SPRING_DATASOURCE_PASSWORD:postgres}
 logging.level.org.springframework.ai.chat.client.advisor=DEBUG
 logging.level.org.springframework.web.servlet.mvc.method.annotation.RequestMappingHandlerMapping=DEBUG
 logging.level.org.springframework.web=DEBUG
 logging.level.org.flywaydb=DEBUG
 logging.level.com.balex.rag.controller=DEBUG
 app.document.chunk-size=200
-#spring.main.allow-circular-references=true
 server.compression.enabled=false
 server.tomcat.connection-timeout=60000
 spring.mvc.async.request-timeout=60000
-#spring.main.web-application-type=reactive
 end.points.users=/users
 end.points.id=/{id}
 end.points.all=/all