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

2025-12-10 11:07:06 +00:00 · 2024-06-22 16:59:15 +08:00
parent 782d4ead9e
commit 32e2c1e39d
5 changed files with 53 additions and 34 deletions
--- a/chat/server/src/main/java/com/tencent/supersonic/chat/server/parser/MultiTurnParser.java
+++ b/chat/server/src/main/java/com/tencent/supersonic/chat/server/parser/MultiTurnParser.java
@@ -1,55 +1,54 @@
 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 {

    private static final Logger keyPipelineLog = LoggerFactory.getLogger("keyPipeline");

    private static final String instruction = ""
-                    + "#Role: You are a data product manager experienced in data requirements.\n"
-                    + "#Task: Your will be provided with current and history questions asked by a user,"
-                    + "along with their mapped schema elements(metric, dimension and value),"
-                    + "please try understanding the semantics and rewrite a question.\n"
-                    + "#Rules: "
-                    + "1.ALWAYS keep relevant entities, metrics, dimensions, values and date ranges. "
-                    + "2.ONLY respond with the rewritten question.\n"
-                    + "#Current Question: %s\n"
-                    + "#Current Mapped Schema: %s\n"
-                    + "#History Question: %s\n"
-                    + "#History Mapped Schema: %s\n"
-                    + "#History SQL: %s\n"
-                    + "#Rewritten Question: ";
+            + "#Role: You are a data product manager experienced in data requirements.\n"
+            + "#Task: Your will be provided with current and history questions asked by a user,"
+            + "along with their mapped schema elements(metric, dimension and value),"
+            + "please try understanding the semantics and rewrite a question.\n"
+            + "#Rules: "
+            + "1.ALWAYS keep relevant entities, metrics, dimensions, values and date ranges. "
+            + "2.ONLY respond with the rewritten question.\n"
+            + "#Current Question: %s\n"
+            + "#Current Mapped Schema: %s\n"
+            + "#History Question: %s\n"
+            + "#History Mapped Schema: %s\n"
+            + "#History SQL: %s\n"
+            + "#Rewritten Question: ";

    @Override
    public void parse(ChatParseContext chatParseContext, ParseResp parseResp) {
@@ -79,13 +78,13 @@ public class MultiTurnParser implements ChatParser {
        String histMapStr = generateSchemaPrompt(lastParseResult.getSelectedParses().get(0).getElementMatches());
        String histSQL = lastParseResult.getSelectedParses().get(0).getSqlInfo().getCorrectS2SQL();
        String rewrittenQuery = rewriteQuery(RewriteContext.builder()
-                        .curtQuestion(currentMapResult.getQueryText())
-                        .histQuestion(lastParseResult.getQueryText())
-                        .curtSchema(curtMapStr)
-                        .histSchema(histMapStr)
-                        .histSQL(histSQL)
-                        .llmConfig(queryReq.getLlmConfig())
-                        .build());
+                .curtQuestion(currentMapResult.getQueryText())
+                .histQuestion(lastParseResult.getQueryText())
+                .curtSchema(curtMapStr)
+                .histSchema(histMapStr)
+                .histSQL(histSQL)
+                .llmConfig(queryReq.getLlmConfig())
+                .build());
        chatParseContext.setQueryText(rewrittenQuery);
        log.info("Last Query: {} Current Query: {}, Rewritten Query: {}",
                lastParseResult.getQueryText(), currentMapResult.getQueryText(), rewrittenQuery);
@@ -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;
--- a/common/pom.xml
+++ b/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>
--- a/headless/chat/src/main/java/com/tencent/supersonic/headless/chat/parser/llm/OnePassSCSqlGenStrategy.java
+++ b/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);
--- a/launchers/standalone/src/main/resources/application-local.yaml
+++ b/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}
+#      api-key: ${OPENAI_API_KEY:demo}
+#  dashscope:
+#    chat-model:
+#      api-key: ${OPENAI_API_KEY:demo}
+#      model-name: qwen-max-1201
+#    embedding-model:
+#      api-key: ${OPENAI_API_KEY:demo}
+
+
--- a/pom.xml
+++ b/pom.xml
@@ -199,6 +199,12 @@
                <groupId>dev.langchain4j</groupId>
                <artifactId>langchain4j-dashscope</artifactId>
                <version>${langchain4j.version}</version>
+                <exclusions>
+                    <exclusion>
+                        <groupId>org.slf4j</groupId>
+                        <artifactId>slf4j-simple</artifactId>
+                    </exclusion>
+                </exclusions>
            </dependency>

            <dependency>