From 574bc2e8dba9d1d11be124f2054ba2243fca125c Mon Sep 17 00:00:00 2001
From: Administrator <15274802129@163.com>
Date: Wed, 03 Sep 2025 13:43:21 +0800
Subject: [PATCH] feat(ai): 优化 prompt 构建逻辑并添加思考内容输出- 在 AliLlmStrategyServiceImpl 中添加了对思考内容的输出- 在 ApiMemberTalkStreamServiceImpl 中优化了 prompt 的构建逻辑 - 在 TestController 中调整了测试用例，使用枚举生成 prompt- 在 TestController 中添加了对思考内容的输出
---
src/main/java/cc/mrbird/febs/ai/strategy/Impl/HsLlmStrategyServiceImpl.java | 99 ++++++++++++++++++++++++++++++++++---------------
1 files changed, 68 insertions(+), 31 deletions(-)
diff --git a/src/main/java/cc/mrbird/febs/ai/strategy/Impl/HsLlmStrategyServiceImpl.java b/src/main/java/cc/mrbird/febs/ai/strategy/Impl/HsLlmStrategyServiceImpl.java
index eb14b9e..4cb18dd 100644
--- a/src/main/java/cc/mrbird/febs/ai/strategy/Impl/HsLlmStrategyServiceImpl.java
+++ b/src/main/java/cc/mrbird/febs/ai/strategy/Impl/HsLlmStrategyServiceImpl.java
@@ -1,29 +1,24 @@
package cc.mrbird.febs.ai.strategy.Impl;
-import cc.mrbird.febs.ai.entity.AiTalkItem;
-import cc.mrbird.febs.ai.res.ai.Report;
-import cc.mrbird.febs.ai.res.memberTalk.ApiMemberTalkStreamVo;
import cc.mrbird.febs.ai.strategy.LlmStrategyService;
+import cc.mrbird.febs.ai.strategy.enumerates.LlmStrategyContextEnum;
import cc.mrbird.febs.ai.strategy.param.LlmStrategyDto;
import cc.mrbird.febs.common.entity.FebsResponse;
import cc.mrbird.febs.common.exception.FebsException;
import cn.hutool.core.collection.CollUtil;
+import cn.hutool.core.util.ObjectUtil;
import cn.hutool.core.util.StrUtil;
-import com.alibaba.dashscope.common.Message;
-import com.alibaba.dashscope.common.Role;
-import com.fasterxml.jackson.core.JsonProcessingException;
-import com.fasterxml.jackson.databind.JsonNode;
import com.volcengine.ark.runtime.model.completion.chat.*;
import com.volcengine.ark.runtime.service.ArkService;
import okhttp3.ConnectionPool;
import okhttp3.Dispatcher;
import org.springframework.stereotype.Component;
-import org.springframework.util.StringUtils;
import reactor.core.publisher.Flux;
import javax.annotation.PostConstruct;
import javax.annotation.PreDestroy;
import java.util.ArrayList;
+import java.util.HashMap;
import java.util.List;
import java.util.concurrent.TimeUnit;
import java.util.stream.Collectors;
@@ -33,7 +28,14 @@
private ArkService service;
+ private static final String ak = "AKLTZTQxZjMyZTUxMWJmNDEyNDkzNWExOGQ3ODllNzhhNmQ";
+ private static final String sk = "TmpFeE1qZ3haREExTW1JeE5HRTBZVGc1WlRRNVlqWXpORGd5TWpsak5HWQ==";
+ private static final String baseUrl = "https://ark.cn-beijing.volces.com/api/v3";
private static final String LinkId = "ep-20250805124033-lhxbf";
+ private static final Double temperature = 0.7;
+ private static final Double topP = 0.9;
+ private static final Integer maxTokens = 2048;
+ private static final Double frequencyPenalty = 0.0;
@PostConstruct
public void init() {
@@ -47,9 +49,9 @@
this.service = ArkService.builder()
.dispatcher(dispatcher)
.connectionPool(connectionPool)
- .baseUrl("https://ark.cn-beijing.volces.com/api/v3")
- .ak("AKLTZTQxZjMyZTUxMWJmNDEyNDkzNWExOGQ3ODllNzhhNmQ")
- .sk("TmpFeE1qZ3haREExTW1JeE5HRTBZVGc1WlRRNVlqWXpORGd5TWpsak5HWQ==")
+ .baseUrl(baseUrl)
+ .ak(ak)
+ .sk(sk)
.build();
}
@@ -96,10 +98,10 @@
.model(LinkId)
.messages(messages)
.stream(false)
- .temperature(0.7) // 降低温度参数，提高确定性，可能提升速度
- .topP(0.9) // 调整topP参数
- .maxTokens(2048) // 减少最大token数
- .frequencyPenalty(0.0)
+ .temperature(temperature) // 降低温度参数，提高确定性，可能提升速度
+ .topP(topP) // 调整topP参数
+ .maxTokens(maxTokens) // 减少最大token数
+ .frequencyPenalty(frequencyPenalty)
.build();
List<ChatCompletionChoice> choices = service.createChatCompletion(chatCompletionRequest).getChoices();
@@ -115,7 +117,7 @@
}
@Override
- public Flux<FebsResponse> llmInvokeStreaming(List<LlmStrategyDto> dto) {
+ public Flux<FebsResponse> llmInvokeStreamingWithThink(List<LlmStrategyDto> dto) {
if (CollUtil.isEmpty(dto)){
throw new FebsException("火山大模型初始化异常");
}
@@ -126,10 +128,10 @@
.messages(messages)
.stream(true)
.thinking(new ChatCompletionRequest.ChatCompletionRequestThinking("enabled"))
- .temperature(0.7)
- .topP(0.9)
- .maxTokens(2048)
- .frequencyPenalty(0.0)
+ .temperature(temperature) // 降低温度参数，提高确定性，可能提升速度
+ .topP(topP) // 调整topP参数
+ .maxTokens(maxTokens) // 减少最大token数
+ .frequencyPenalty(frequencyPenalty)
.build();
return Flux.from(service.streamChatCompletion(chatCompletionRequest))
@@ -144,21 +146,56 @@
}
ChatMessage message = choice.getMessage();
- ApiMemberTalkStreamVo apiMemberTalkStreamVo = new ApiMemberTalkStreamVo();
- // 处理 reasoning content
- String reasoningContent = message.getReasoningContent();
- if (StrUtil.isNotEmpty(reasoningContent)) {
- apiMemberTalkStreamVo.setReasoningContent(reasoningContent);
+
+ HashMap<String, String> stringStringHashMap = new HashMap<>();
+ if (ObjectUtil.isNotEmpty(message.getReasoningContent())) {
+ stringStringHashMap.put(LlmStrategyContextEnum.THINK.name(),message.getReasoningContent().toString());
+ }
+ if (ObjectUtil.isNotEmpty(message.getContent())) {
+ stringStringHashMap.put(LlmStrategyContextEnum.CONTENT.name(),message.getContent().toString());
+ }
+ return new FebsResponse().success().data(stringStringHashMap);
+ })
+ .onErrorResume(throwable -> {
+ throw new FebsException(StrUtil.format("火山大模型流式调用AI服务失:{}",throwable));
+ });
+ }
+
+ @Override
+ public Flux<FebsResponse> llmInvokeStreamingNoThink(List<LlmStrategyDto> dto) {
+ if (CollUtil.isEmpty(dto)){
+ throw new FebsException("火山大模型初始化异常");
+ }
+ List<ChatMessage> messages = getMessages(dto);
+
+ ChatCompletionRequest chatCompletionRequest = ChatCompletionRequest.builder()
+ .model(LinkId)
+ .messages(messages)
+ .stream(true)
+ .temperature(temperature) // 降低温度参数，提高确定性，可能提升速度
+ .topP(topP) // 调整topP参数
+ .maxTokens(maxTokens) // 减少最大token数
+ .frequencyPenalty(frequencyPenalty)
+ .build();
+
+ return Flux.from(service.streamChatCompletion(chatCompletionRequest))
+ .map(response -> {
+ if (response == null || response.getChoices() == null || response.getChoices().isEmpty()) {
+ return new FebsResponse().success().data("未响应，请重试");
}
- // 安全处理 content
- String content = "";
- if (message.getContent() != null) {
- content = message.getContent().toString();
+ ChatCompletionChoice choice = response.getChoices().get(0);
+ if (choice == null || choice.getMessage() == null) {
+ return new FebsResponse().success().data("END");
}
- apiMemberTalkStreamVo.setContent(content);
- return new FebsResponse().success().data(apiMemberTalkStreamVo);
+ ChatMessage message = choice.getMessage();
+
+ HashMap<String, String> stringStringHashMap = new HashMap<>();
+ if (ObjectUtil.isNotEmpty(message.getContent())) {
+ stringStringHashMap.put(LlmStrategyContextEnum.CONTENT.name(),message.getContent().toString());
+ }
+ return new FebsResponse().success().data(stringStringHashMap);
})
.onErrorResume(throwable -> {
throw new FebsException(StrUtil.format("火山大模型流式调用AI服务失:{}",throwable));
--
Gitblit v1.9.1