(improvement)(common) Fixed the compatibility issue with qwen. (#1193)

chat/server/src/main/java/com/tencent/supersonic/chat/server/parser/MultiTurnParser.java

@@ -1,36 +1,35 @@
 package com.tencent.supersonic.chat.server.parser;
 
+import static com.tencent.supersonic.chat.server.parser.ParserConfig.PARSER_MULTI_TURN_ENABLE;
+
 import com.tencent.supersonic.chat.server.agent.MultiTurnConfig;
 import com.tencent.supersonic.chat.server.persistence.repository.ChatQueryRepository;
 import com.tencent.supersonic.chat.server.pojo.ChatParseContext;
 import com.tencent.supersonic.chat.server.util.QueryReqConverter;
-import com.tencent.supersonic.common.util.ContextUtils;
 import com.tencent.supersonic.common.config.LLMConfig;
+import com.tencent.supersonic.common.util.ContextUtils;
+import com.tencent.supersonic.common.util.S2ChatModelProvider;
 import com.tencent.supersonic.headless.api.pojo.SchemaElementMatch;
 import com.tencent.supersonic.headless.api.pojo.SchemaElementType;
 import com.tencent.supersonic.headless.api.pojo.request.QueryReq;
 import com.tencent.supersonic.headless.api.pojo.response.MapResp;
 import com.tencent.supersonic.headless.api.pojo.response.ParseResp;
-import com.tencent.supersonic.common.util.S2ChatModelProvider;
 import com.tencent.supersonic.headless.server.service.ChatQueryService;
 import dev.langchain4j.data.message.AiMessage;
 import dev.langchain4j.model.chat.ChatLanguageModel;
 import dev.langchain4j.model.input.Prompt;
 import dev.langchain4j.model.input.PromptTemplate;
 import dev.langchain4j.model.output.Response;
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.List;
+import java.util.stream.Collectors;
 import lombok.Builder;
 import lombok.Data;
 import lombok.extern.slf4j.Slf4j;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
-import java.util.ArrayList;
-import java.util.List;
-import java.util.stream.Collectors;
-import java.util.Collections;
-
-import static com.tencent.supersonic.chat.server.parser.ParserConfig.PARSER_MULTI_TURN_ENABLE;
-
 @Slf4j
 public class MultiTurnParser implements ChatParser {
 
@@ -98,7 +97,7 @@ public class MultiTurnParser implements ChatParser {
         keyPipelineLog.info("MultiTurnParser reqPrompt:{}", promptStr);
 
         ChatLanguageModel chatLanguageModel = S2ChatModelProvider.provide(context.getLlmConfig());
-        Response<AiMessage> response = chatLanguageModel.generate(prompt.toSystemMessage());
+        Response<AiMessage> response = chatLanguageModel.generate(prompt.toUserMessage());
 
         String result = response.content().text();
         keyPipelineLog.info("MultiTurnParser modelResp:{}", result);
@@ -144,6 +143,7 @@ public class MultiTurnParser implements ChatParser {
     @Data
     @Builder
     public static class RewriteContext {
+
         private String curtQuestion;
         private String histQuestion;
         private String curtSchema;

common/pom.xml

@@ -228,6 +228,12 @@
         <dependency>
             <groupId>dev.langchain4j</groupId>
             <artifactId>langchain4j-dashscope</artifactId>
+            <exclusions>
+                <exclusion>
+                    <groupId>org.slf4j</groupId>
+                    <artifactId>slf4j-simple</artifactId>
+                </exclusion>
+            </exclusions>
         </dependency>
         <!--langchain4j-->
         <dependency>

headless/chat/src/main/java/com/tencent/supersonic/headless/chat/parser/llm/OnePassSCSqlGenStrategy.java

@@ -39,9 +39,9 @@ public class OnePassSCSqlGenStrategy extends SqlGenStrategy {
         //3.perform multiple self-consistency inferences parallelly
         Map<Prompt, String> prompt2Output = new ConcurrentHashMap<>();
         prompt2Exemplar.keySet().parallelStream().forEach(prompt -> {
-                    keyPipelineLog.info("OnePassSCSqlGenStrategy reqPrompt:\n{}", prompt.toSystemMessage());
+                    keyPipelineLog.info("OnePassSCSqlGenStrategy reqPrompt:\n{}", prompt.toUserMessage());
                     ChatLanguageModel chatLanguageModel = getChatLanguageModel(llmReq.getLlmConfig());
-                    Response<AiMessage> response = chatLanguageModel.generate(prompt.toSystemMessage());
+                    Response<AiMessage> response = chatLanguageModel.generate(prompt.toUserMessage());
                     String result = response.content().text();
                     prompt2Output.put(prompt, result);
                     keyPipelineLog.info("OnePassSCSqlGenStrategy modelResp:\n{}", result);

launchers/standalone/src/main/resources/application-local.yaml

@@ -99,7 +99,14 @@ langchain4j:
       model-name: ${OPENAI_MODEL_NAME:gpt-3.5-turbo}
       temperature: ${OPENAI_TEMPERATURE:0.0}
       timeout: ${OPENAI_TIMEOUT:PT60S}
-# java.lang.RuntimeException: dev.ai4j.openai4j.OpenAiHttpException: Too many requests
 #    embedding-model:
 #      base-url: ${OPENAI_API_BASE:https://api.openai.com/v1}
 #      api-key: ${OPENAI_API_KEY:demo}
+#  dashscope:
+#    chat-model:
+#      api-key: ${OPENAI_API_KEY:demo}
+#      model-name: qwen-max-1201
+#    embedding-model:
+#      api-key: ${OPENAI_API_KEY:demo}
+
+

pom.xml

@@ -199,6 +199,12 @@
                 <groupId>dev.langchain4j</groupId>
                 <artifactId>langchain4j-dashscope</artifactId>
                 <version>${langchain4j.version}</version>
+                <exclusions>
+                    <exclusion>
+                        <groupId>org.slf4j</groupId>
+                        <artifactId>slf4j-simple</artifactId>
+                    </exclusion>
+                </exclusions>
             </dependency>
 
             <dependency>