topdog/supersonic
1
0
Fork 0
mirror of https://github.com/tencentmusic/supersonic.git synced 2025-12-11 12:07:42 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

[feature][headless-chat]Introduce ChatApp to support more flexible chat model config.#1739

Browse Source
This commit is contained in:
jerryjzhang
2024-10-12 11:51:37 +08:00
parent 4408bf4325
commit 2717a1603c
25 changed files with 220 additions and 175 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
headless/api/src/main/java/com/tencent/supersonic/headless/api/pojo/request/QueryNLReq.java
View File

@@ -3,6 +3,7 @@ package com.tencent.supersonic.headless.api.pojo.request;
import com.google.common.collect.Lists;
import com.google.common.collect.Sets;
import com.tencent.supersonic.auth.api.authentication.pojo.User;
import com.tencent.supersonic.common.pojo.ChatApp;
import com.tencent.supersonic.common.pojo.ChatModelConfig;
import com.tencent.supersonic.common.pojo.Text2SQLExemplar;
import com.tencent.supersonic.common.pojo.enums.Text2SQLType;
@@ -13,6 +14,7 @@ import com.tencent.supersonic.headless.api.pojo.enums.MapModeEnum;
import lombok.Data;
import java.util.List;
import java.util.Map;
import java.util.Set;
@Data
@@ -26,8 +28,7 @@ public class QueryNLReq {
private MapModeEnum mapModeEnum = MapModeEnum.STRICT;
private SchemaMapInfo mapInfo = new SchemaMapInfo();
private QueryDataType queryDataType = QueryDataType.ALL;
private ChatModelConfig modelConfig;
private String customPrompt;
private Map<String, ChatApp> chatAppConfig;
private List<Text2SQLExemplar> dynamicExemplars = Lists.newArrayList();
private SemanticParseInfo contextParseInfo;
}

8
headless/chat/src/main/java/com/tencent/supersonic/headless/chat/ChatQueryContext.java
View File

@@ -2,6 +2,7 @@ package com.tencent.supersonic.headless.chat;
import com.fasterxml.jackson.annotation.JsonIgnore;
import com.tencent.supersonic.auth.api.authentication.pojo.User;
import com.tencent.supersonic.common.pojo.ChatApp;
import com.tencent.supersonic.common.pojo.ChatModelConfig;
import com.tencent.supersonic.common.pojo.Text2SQLExemplar;
import com.tencent.supersonic.common.pojo.enums.Text2SQLType;
@@ -47,13 +48,14 @@ public class ChatQueryContext {
private SchemaMapInfo mapInfo = new SchemaMapInfo();
private SemanticParseInfo contextParseInfo;
private MapModeEnum mapModeEnum = MapModeEnum.STRICT;
private QueryDataType queryDataType = QueryDataType.ALL;
@JsonIgnore
private SemanticSchema semanticSchema;
@JsonIgnore
private ChatWorkflowState chatWorkflowState;
private QueryDataType queryDataType = QueryDataType.ALL;
private ChatModelConfig modelConfig;
private String customPrompt;
@JsonIgnore
private Map<String, ChatApp> chatAppConfig;
@JsonIgnore
private List<Text2SQLExemplar> dynamicExemplars;
public List<SemanticQuery> getCandidateQueries() {

25
headless/chat/src/main/java/com/tencent/supersonic/headless/chat/corrector/LLMSqlCorrector.java
View File

@@ -1,5 +1,7 @@
package com.tencent.supersonic.headless.chat.corrector;
import com.tencent.supersonic.common.pojo.ChatApp;
import com.tencent.supersonic.common.util.ChatAppManager;
import com.tencent.supersonic.headless.api.pojo.SemanticParseInfo;
import com.tencent.supersonic.headless.chat.ChatQueryContext;
import dev.langchain4j.model.chat.ChatLanguageModel;
@@ -23,11 +25,11 @@ public class LLMSqlCorrector extends BaseSemanticCorrector {
private static final Logger keyPipelineLog = LoggerFactory.getLogger("keyPipeline");
public static final String INSTRUCTION = ""
public static final String APP_KEY = "S2SQL_CORRECTOR";
private static final String INSTRUCTION = ""
+ "\n#Role: You are a senior data engineer experienced in writing SQL."
+ "\n#Task: Your will be provided with a user question and the SQL written by a junior engineer,"
+ "please take a review and help correct it if necessary."
+ "\n#Rules: "
+ "please take a review and help correct it if necessary." + "\n#Rules: "
+ "\n1.ALWAYS follow the output format: `opinion=(POSITIVE|NEGATIVE),sql=(corrected sql if NEGATIVE; empty string if POSITIVE)`."
+ "\n2.NO NEED to check date filters as the junior engineer seldom makes mistakes in this regard."
+ "\n3.DO NOT miss the AGGREGATE operator of metrics, always add it as needed."
@@ -36,6 +38,11 @@ public class LLMSqlCorrector extends BaseSemanticCorrector {
+ "\n6.ALWAYS translate alias created by `AS` command to the same language as the `#Question`."
+ "\n#Question:{{question}} #InputSQL:{{sql}} #Response:";
public LLMSqlCorrector() {
ChatAppManager.register(ChatApp.builder().key(APP_KEY).prompt(INSTRUCTION).name("语义SQL修正")
.description("").enable(false).build());
}
@Data
@ToString
static class SemanticSql {
@@ -52,14 +59,16 @@ public class LLMSqlCorrector extends BaseSemanticCorrector {
@Override
public void doCorrect(ChatQueryContext chatQueryContext, SemanticParseInfo semanticParseInfo) {
if (!chatQueryContext.getText2SQLType().enableLLM()) {
ChatApp chatApp = chatQueryContext.getChatAppConfig().get(APP_KEY);
if (!chatQueryContext.getText2SQLType().enableLLM() || !chatApp.isEnable()) {
return;
}
ChatLanguageModel chatLanguageModel = ModelProvider.getChatModel(chatQueryContext.getModelConfig());
ChatLanguageModel chatLanguageModel = ModelProvider.getChatModel(chatApp.getChatModelConfig());
SemanticSqlExtractor extractor =
AiServices.create(SemanticSqlExtractor.class, chatLanguageModel);
Prompt prompt = generatePrompt(chatQueryContext.getQueryText(), semanticParseInfo);
Prompt prompt = generatePrompt(chatQueryContext.getQueryText(), semanticParseInfo,
chatApp.getPrompt());
keyPipelineLog.info("LLMSqlCorrector reqPrompt:\n{}", prompt.text());
SemanticSql s2Sql = extractor.generateSemanticSql(prompt.toUserMessage().singleText());
keyPipelineLog.info("LLMSqlCorrector modelResp:\n{}", s2Sql);
@@ -68,12 +77,12 @@ public class LLMSqlCorrector extends BaseSemanticCorrector {
}
}
private Prompt generatePrompt(String queryText, SemanticParseInfo semanticParseInfo) {
private Prompt generatePrompt(String queryText, SemanticParseInfo semanticParseInfo,
String promptTemplate) {
Map<String, Object> variable = new HashMap<>();
variable.put("question", queryText);
variable.put("sql", semanticParseInfo.getSqlInfo().getCorrectedS2SQL());
String promptTemplate = INSTRUCTION;
return PromptTemplate.from(promptTemplate).apply(variable);
}
}

3
headless/chat/src/main/java/com/tencent/supersonic/headless/chat/parser/llm/LLMRequestService.java
View File

@@ -74,8 +74,7 @@ public class LLMRequestService {
llmReq.setTerms(getMappedTerms(queryCtx, dataSetId));
llmReq.setSqlGenType(
LLMReq.SqlGenType.valueOf(parserConfig.getParameterValue(PARSER_STRATEGY_TYPE)));
llmReq.setModelConfig(queryCtx.getModelConfig());
llmReq.setCustomPrompt(queryCtx.getCustomPrompt());
llmReq.setChatAppConfig(queryCtx.getChatAppConfig());
llmReq.setDynamicExemplars(queryCtx.getDynamicExemplars());
return llmReq;

7
headless/chat/src/main/java/com/tencent/supersonic/headless/chat/parser/llm/LLMSqlParser.java
View File

@@ -1,5 +1,6 @@
package com.tencent.supersonic.headless.chat.parser.llm;
import com.tencent.supersonic.common.pojo.ChatModelConfig;
import com.tencent.supersonic.common.util.ContextUtils;
import com.tencent.supersonic.headless.chat.ChatQueryContext;
import com.tencent.supersonic.headless.chat.parser.SemanticParser;
@@ -69,10 +70,12 @@ public class LLMSqlParser implements SemanticParser {
} catch (Exception e) {
log.error("currentRetryRound:{}, runText2SQL failed", currentRetry, e);
}
Double temperature = llmReq.getModelConfig().getTemperature();
ChatModelConfig chatModelConfig = llmReq.getChatAppConfig()
.get(OnePassSCSqlGenStrategy.APP_KEY).getChatModelConfig();
Double temperature = chatModelConfig.getTemperature();
if (temperature == 0) {
// 报错时增加随机性，减少无效重试
llmReq.getModelConfig().setTemperature(0.5);
chatModelConfig.setTemperature(0.5);
}
currentRetry++;
}

26
headless/chat/src/main/java/com/tencent/supersonic/headless/chat/parser/llm/OnePassSCSqlGenStrategy.java
View File

@@ -1,7 +1,9 @@
package com.tencent.supersonic.headless.chat.parser.llm;
import com.google.common.collect.Lists;
import com.tencent.supersonic.common.pojo.ChatApp;
import com.tencent.supersonic.common.pojo.Text2SQLExemplar;
import com.tencent.supersonic.common.util.ChatAppManager;
import com.tencent.supersonic.headless.chat.query.llm.s2sql.LLMReq;
import com.tencent.supersonic.headless.chat.query.llm.s2sql.LLMResp;
import dev.langchain4j.model.chat.ChatLanguageModel;
@@ -11,7 +13,6 @@ import dev.langchain4j.model.output.structured.Description;
import dev.langchain4j.service.AiServices;
import lombok.Data;
import lombok.extern.slf4j.Slf4j;
import org.apache.commons.lang3.StringUtils;
import org.apache.commons.lang3.tuple.Pair;
import org.springframework.stereotype.Service;
@@ -24,6 +25,7 @@ import java.util.concurrent.ConcurrentHashMap;
@Slf4j
public class OnePassSCSqlGenStrategy extends SqlGenStrategy {
public static final String APP_KEY = "S2SQL_PARSER";
public static final String INSTRUCTION = ""
+ "\n#Role: You are a data analyst experienced in SQL languages."
+ "\n#Task: You will be provided with a natural language question asked by users,"
@@ -40,6 +42,11 @@ public class OnePassSCSqlGenStrategy extends SqlGenStrategy {
+ "\n#Exemplars: {{exemplar}}"
+ "\n#Question: Question:{{question}},Schema:{{schema}},SideInfo:{{information}}";
public OnePassSCSqlGenStrategy() {
ChatAppManager.register(ChatApp.builder().key(APP_KEY).prompt(INSTRUCTION).name("语义SQL解析")
.description("通过大模型做语义解析生成S2SQL").enable(true).build());
}
@Data
static class SemanticSql {
@Description("thought or remarks to tell users about the sql, make it short.")
@@ -62,15 +69,17 @@ public class OnePassSCSqlGenStrategy extends SqlGenStrategy {
List<List<Text2SQLExemplar>> exemplarsList = promptHelper.getFewShotExemplars(llmReq);
// 2.generate sql generation prompt for each self-consistency inference
ChatApp chatApp = llmReq.getChatAppConfig().get(APP_KEY);
ChatLanguageModel chatLanguageModel = getChatLanguageModel(chatApp.getChatModelConfig());
SemanticSqlExtractor extractor =
AiServices.create(SemanticSqlExtractor.class, chatLanguageModel);
Map<Prompt, List<Text2SQLExemplar>> prompt2Exemplar = new HashMap<>();
for (List<Text2SQLExemplar> exemplars : exemplarsList) {
llmReq.setDynamicExemplars(exemplars);
Prompt prompt = generatePrompt(llmReq, llmResp);
Prompt prompt = generatePrompt(llmReq, llmResp, chatApp);
prompt2Exemplar.put(prompt, exemplars);
}
ChatLanguageModel chatLanguageModel = getChatLanguageModel(llmReq.getModelConfig());
SemanticSqlExtractor extractor =
AiServices.create(SemanticSqlExtractor.class, chatLanguageModel);
// 3.perform multiple self-consistency inferences parallelly
Map<String, Prompt> output2Prompt = new ConcurrentHashMap<>();
@@ -92,7 +101,7 @@ public class OnePassSCSqlGenStrategy extends SqlGenStrategy {
return llmResp;
}
private Prompt generatePrompt(LLMReq llmReq, LLMResp llmResp) {
private Prompt generatePrompt(LLMReq llmReq, LLMResp llmResp, ChatApp chatApp) {
StringBuilder exemplars = new StringBuilder();
for (Text2SQLExemplar exemplar : llmReq.getDynamicExemplars()) {
String exemplarStr = String.format("\nQuestion:%s,Schema:%s,SideInfo:%s,SQL:%s",
@@ -112,10 +121,7 @@ public class OnePassSCSqlGenStrategy extends SqlGenStrategy {
variable.put("information", sideInformation);
// use custom prompt template if provided.
String promptTemplate = INSTRUCTION;
if (StringUtils.isNotBlank(llmReq.getCustomPrompt())) {
promptTemplate = llmReq.getCustomPrompt();
}
String promptTemplate = chatApp.getPrompt();
return PromptTemplate.from(promptTemplate).apply(variable);
}

4
headless/chat/src/main/java/com/tencent/supersonic/headless/chat/query/llm/s2sql/LLMReq.java
View File

@@ -2,6 +2,7 @@ package com.tencent.supersonic.headless.chat.query.llm.s2sql;
import com.fasterxml.jackson.annotation.JsonValue;
import com.google.common.collect.Lists;
import com.tencent.supersonic.common.pojo.ChatApp;
import com.tencent.supersonic.common.pojo.ChatModelConfig;
import com.tencent.supersonic.common.pojo.Text2SQLExemplar;
import com.tencent.supersonic.headless.api.pojo.SchemaElement;
@@ -10,6 +11,7 @@ import org.apache.commons.collections4.CollectionUtils;
import java.util.ArrayList;
import java.util.List;
import java.util.Map;
import java.util.Objects;
import java.util.stream.Collectors;
@@ -21,7 +23,7 @@ public class LLMReq {
private String currentDate;
private String priorExts;
private SqlGenType sqlGenType;
private ChatModelConfig modelConfig;
private Map<String, ChatApp> chatAppConfig;
private String customPrompt;
private List<Text2SQLExemplar> dynamicExemplars;