refactor(module-llm):优化数据集处理和定时任务执行频率
- 重构 AsyncDataSetService 中的 JsonFileWrite 方法,提高数据集处理效率 - 修改 DatasetTaskSyncService 中的定时任务执行频率,从每 30秒调整为每 15秒
This commit is contained in:
parent
b5195e1113
commit
f7a501cc7d
@ -34,29 +34,27 @@ public class AsyncDataSetService {
|
||||
|
||||
@Async
|
||||
public void JsonFileWrite(DatasetDO datasetDO,List<DatasetQuestionRespVO> datasetQuestionList) {
|
||||
StringBuilder sb = new StringBuilder();
|
||||
List<AigcDatasetVo> aigcDatasetVoList = new ArrayList<>();
|
||||
datasetQuestionList.forEach(dataSource -> {
|
||||
List<ModelCompletionsReqVO.ModelCompletionsMessage> messages = new ArrayList<>();
|
||||
ModelCompletionsReqVO.ModelCompletionsMessage questionMessage = new ModelCompletionsReqVO.ModelCompletionsMessage();
|
||||
questionMessage.setRole("user");
|
||||
questionMessage.setContent(StringUtils.isNotBlank(dataSource.getQuestion())?dataSource.getQuestion():"");
|
||||
messages.add(questionMessage);
|
||||
ModelCompletionsReqVO.ModelCompletionsMessage answerMessage = new ModelCompletionsReqVO.ModelCompletionsMessage();
|
||||
answerMessage.setRole("assistant");
|
||||
AigcDatasetVo aigcDatasetVo = new AigcDatasetVo();
|
||||
aigcDatasetVo.setInstruction(StringUtils.isNotBlank(dataSource.getSystem())?dataSource.getSystem():"");
|
||||
aigcDatasetVo.setInput(StringUtils.isNotBlank(dataSource.getQuestion())?dataSource.getQuestion():"");
|
||||
if (!CollectionUtils.isAnyEmpty(dataSource.getDatasetAnswerRespVO())){
|
||||
answerMessage.setContent(StringUtils.isNotBlank(dataSource.getDatasetAnswerRespVO().get(0).getAnswer())?dataSource.getDatasetAnswerRespVO().get(0).getAnswer():"");
|
||||
aigcDatasetVo.setOutput(StringUtils.isNotBlank(dataSource.getDatasetAnswerRespVO().get(0).getAnswer())?dataSource.getDatasetAnswerRespVO().get(0).getAnswer():"");
|
||||
}else {
|
||||
answerMessage.setContent("");
|
||||
aigcDatasetVo.setOutput("");
|
||||
}
|
||||
messages.add(answerMessage);
|
||||
JSONObject jsonObject = new JSONObject();
|
||||
jsonObject.put("messages", messages);
|
||||
sb.append(jsonObject.toString()).append("\n");
|
||||
|
||||
aigcDatasetVoList.add(aigcDatasetVo);
|
||||
});
|
||||
ObjectMapper mapper = new ObjectMapper();
|
||||
try (InputStream inputStream = new ByteArrayInputStream(sb.toString().getBytes())){
|
||||
AigcDatasetFileRespV0 aigcDatasetFileRespV0 = trainHttpService.AigcUploadFile(new HashMap<>(), inputStream, datasetDO.getDatasetName() + datasetDO.getId() + ".json");
|
||||
StringBuilder sb = new StringBuilder();
|
||||
try {
|
||||
for (AigcDatasetVo aigcDatasetVo : aigcDatasetVoList) {
|
||||
String json = mapper.writeValueAsString(aigcDatasetVo);
|
||||
sb.append(json).append("\n"); // 每个 JSON 对象后换行
|
||||
}
|
||||
InputStream inputStream = new ByteArrayInputStream(sb.toString().getBytes());
|
||||
AigcDatasetFileRespV0 aigcDatasetFileRespV0 = trainHttpService.AigcUploadFile(new HashMap<>(), inputStream, datasetDO.getDatasetName() + "new"+datasetDO.getId() + ".json");
|
||||
if (aigcDatasetFileRespV0 != null){
|
||||
datasetMapper.setJobid(datasetDO.getId(),aigcDatasetFileRespV0.getFileId());
|
||||
|
||||
|
@ -23,7 +23,7 @@ public class DatasetTaskSyncService {
|
||||
@Resource
|
||||
private AsyncDataSetService dataSetService;
|
||||
|
||||
@Scheduled(cron ="0/30 * * * * ?")
|
||||
@Scheduled(cron ="0/15 * * * * ?")
|
||||
public void syncDatasetAigcTask() {
|
||||
log.info("[syncDatasetAigcTask][开始同步任务]");
|
||||
// 查询已经标注完成的数据集
|
||||
|
Loading…
x
Reference in New Issue
Block a user