perf: 知识库优化

8239c584 · archer · 456686f3 · 8239c584 · 8239c584 · 8239c584
Commit 8239c584 authored Mar 31, 2023 by archer
12 changed files
--- a/src/api/model.ts
+++ b/src/api/model.ts
 import { GET, POST, DELETE, PUT } from './request';
-import type { ModelSchema, ModelDataSchema } from '@/types/mongoSchema';
+import type { ModelSchema, ModelDataSchema, ModelSplitDataSchema } from '@/types/mongoSchema';
 import { ModelUpdateParams } from '@/types/model';
 import { TrainingItemType } from '../types/training';
-import { PagingData } from '@/types';
 import { RequestPaging } from '../types/index';
 import { Obj2Query } from '@/utils/tools';

@@ -39,6 +38,9 @@ type GetModelDataListProps = RequestPaging & {
 export const getModelDataList = (props: GetModelDataListProps) =>
  GET(`/model/data/getModelData?${Obj2Query(props)}`);

+export const getModelSplitDataList = (modelId: string) =>
+  GET<ModelSplitDataSchema[]>(`/model/data/getSplitData?modelId=${modelId}`);
+
 export const postModelDataInput = (data: {
  modelId: string;
  data: { text: ModelDataSchema['text']; q: ModelDataSchema['q'] }[];

--- a/src/constants/common.ts
+++ b/src/constants/common.ts
@@ -17,30 +17,24 @@ export const introPage = `
 4. 进入模型页，创建一个模型，建议直接用 ChatGPT。    
 5. 在模型列表点击【对话】，即可使用 API 进行聊天。  

-### 模型配置
+### 定制 prompt

-1. **提示语**：会在每个对话框的第一句自动加入，用于限定该模型的对话内容。  
+1. 进入模型编辑页  
+2. 调整温度和提示词  
+3. 使用该模型对话。每次对话时，提示词和温度都会自动注入，方便管理个人的模型。建议把自己日常经常需要使用的 5~10 个方向预设好。

+### 知识库

-2. **单句最大长度**：每个聊天，单次输入内容的最大长度。  
+1. 创建模型时选择【知识库】  
+2. 进入模型编辑页  
+3. 导入数据，可以选择手动导入，或者选择文件导入。文件导入会自动调用 chatGPT 理解文件内容，并生成知识库。  
+4. 使用该模型对话。  

-
-3. **上下文最大长度**：每个聊天，最多的轮数除以2，建议设置为偶数。可以持续聊天，但是旧的聊天内容会被截断，AI 就不会知道被截取的内容。 
-例如：上下文最大长度为6。在第 4 轮对话时，第一轮对话的内容不会被计入。
-
-4. **过期时间**：生成对话框后，这个对话框多久过期。  
-
-5. **聊天最大加载次数**：单个对话框最多被加载几次，设置为-1代表不限制，正数代表只能加载 n 次，防止被盗刷。  
-
-### 对话框介绍
-
-1. 每个对话框以 chatId 作为标识。  
-2. 每次点击【对话】，都会生成新的对话框，无法回到旧的对话框。对话框内刷新，会恢复对话内容。  
-3. 直接分享对话框（网页）的链接给朋友，会共享同一个对话内容。但是！！！千万不要两个人同时用一个链接，会串味，还没解决这个问题。  
-4. 如果想分享一个纯的对话框，请点击侧边栏的分享按键。例如：  
+注意：使用知识库模型对话时，tokens 消耗会加快。  

 ### 其他问题
 还有其他问题，可以加我 wx: YNyiqi，拉个交流群大家一起聊聊。
+
 `;

 export const chatProblem = `
@@ -58,10 +52,9 @@ export const chatProblem = `
 `;

 export const versionIntro = `
-## Fast GPT V2.0
+## Fast GPT V2.2
+* 定制知识库：创建模型时可以选择【知识库】模型, 可以手动导入知识点或者直接导入一个文件自动学习。
 * 删除和复制功能：点击对话头像，可以选择复制或删除该条内容。
-* 优化记账模式： 不再根据文本长度进行记账，而是根据实际消耗 tokens 数量进行记账。
-* 文本 QA 拆分: 可以在[数据]模块，使用 QA 拆分功能，粘贴文字或者选择文件均可以实现自动生成 QA。可以一键导出，用于微调模型。
 `;

 export const shareHint = `

--- a/src/hooks/usePagination.tsx
+++ b/src/hooks/usePagination.tsx
+import { useState, useCallback, useMemo } from 'react';
+import type { PagingData } from '../types/index';
+import { IconButton, Flex, Box } from '@chakra-ui/react';
+import { ArrowBackIcon, ArrowForwardIcon } from '@chakra-ui/icons';
+import { useQuery, useMutation } from '@tanstack/react-query';
+import { useToast } from './useToast';
+
+export const usePagination = <T = any,>({
+  api,
+  pageSize = 10,
+  params = {}
+}: {
+  api: (data: any) => any;
+  pageSize?: number;
+  params?: Record<string, any>;
+}) => {
+  const { toast } = useToast();
+  const [pageNum, setPageNum] = useState(1);
+  const [total, setTotal] = useState(0);
+  const maxPage = useMemo(() => Math.ceil(total / pageSize), [pageSize, total]);
+
+  const {
+    mutate,
+    data = [],
+    isLoading
+  } = useMutation({
+    mutationFn: async (num: number = pageNum) => {
+      try {
+        const res: PagingData<T> = await api({
+          pageNum: num,
+          pageSize,
+          ...params
+        });
+        setPageNum(num);
+        setTotal(res.total);
+        return res.data;
+      } catch (error: any) {
+        toast({
+          title: error?.message || '获取数据异常',
+          status: 'error'
+        });
+        console.log(error);
+      }
+    }
+  });
+
+  useQuery(['init'], () => {
+    mutate(1);
+    return null;
+  });
+
+  const Pagination = useCallback(() => {
+    return (
+      <Flex alignItems={'center'} justifyContent={'end'}>
+        <IconButton
+          isDisabled={pageNum === 1}
+          icon={<ArrowBackIcon />}
+          aria-label={'left'}
+          size={'sm'}
+          onClick={() => mutate(pageNum - 1)}
+        />
+        <Box mx={2}>
+          {pageNum}/{maxPage}
+        </Box>
+        <IconButton
+          isDisabled={pageNum === maxPage}
+          icon={<ArrowForwardIcon />}
+          aria-label={'left'}
+          size={'sm'}
+          onClick={() => mutate(pageNum + 1)}
+        />
+      </Flex>
+    );
+  }, [maxPage, mutate, pageNum]);
+
+  return {
+    pageNum,
+    pageSize,
+    total,
+    data,
+    isLoading,
+    Pagination,
+    getData: mutate
+  };
+};
--- a/src/pages/api/chat/chatGpt.ts
+++ b/src/pages/api/chat/chatGpt.ts
@@ -40,6 +40,7 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse)
    }

    await connectToDatabase();
+    let startTime = Date.now();

    const { chat, userApiKey, systemKey, userId } = await authChat(chatId, authorization);

@@ -81,7 +82,6 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse)

    // 获取 chatAPI
    const chatAPI = getOpenAIApi(userApiKey || systemKey);
-    let startTime = Date.now();
    // 发出请求
    const chatResponse = await chatAPI.createChatCompletion(
      {

--- a/src/pages/api/chat/vectorGpt.ts
+++ b/src/pages/api/chat/vectorGpt.ts
@@ -48,6 +48,7 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse)

    await connectToDatabase();
    const redis = await connectRedis();
+    let startTime = Date.now();

    const { chat, userApiKey, systemKey, userId } = await authChat(chatId, authorization);

@@ -83,17 +84,22 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse)
    const redisData: any[] = await redis.sendCommand([
      'FT.SEARCH',
      `idx:${VecModelDataIndex}`,
-      `@modelId:{${String(chat.modelId._id)}} @vector:[VECTOR_RANGE 0.2 $blob]`,
+      `@modelId:{${String(
+        chat.modelId._id
+      )}} @vector:[VECTOR_RANGE 0.15 $blob]=>{$YIELD_DISTANCE_AS: score}`,
      // `@modelId:{${String(chat.modelId._id)}}=>[KNN 10 @vector $blob AS score]`,
      'RETURN',
      '1',
      'dataId',
-      // 'SORTBY',
-      // 'score',
+      'SORTBY',
+      'score',
      'PARAMS',
      '2',
      'blob',
      binary,
+      'LIMIT',
+      '0',
+      '20',
      'DIALECT',
      '2'
    ]);
@@ -117,8 +123,13 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse)
        [2, 4, 6, 8, 10, 12, 14, 16, 18, 20].map((i) => {
          if (!redisData[i] || !redisData[i][1]) return '';
          return ModelData.findById(redisData[i][1])
-            .select('text')
-            .then((res) => res?.text || '');
+            .select('text q')
+            .then((res) => {
+              if (!res) return '';
+              const questions = res.q.map((item) => item.text).join(' ');
+              const answer = res.text;
+              return `${questions} ${answer}`;
+            });
        })
      )
    ).filter((item) => item);
@@ -128,7 +139,7 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse)

    prompts.unshift({
      obj: 'SYSTEM',
-      value: `请根据下面的知识回答问题： ${systemPrompt}`
+      value: `根据下面的知识回答问题： ${systemPrompt}`
    });

    // 控制在 tokens 数量，防止超出
@@ -150,7 +161,6 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse)
    // 计算温度
    const temperature = modelConstantsData.maxTemperature * (model.temperature / 10);

-    let startTime = Date.now();
    // 发出请求
    const chatResponse = await chatAPI.createChatCompletion(
      {

--- a/src/pages/api/model/data/getSplitData.ts
+++ b/src/pages/api/model/data/getSplitData.ts
+import type { NextApiRequest, NextApiResponse } from 'next';
+import { jsonRes } from '@/service/response';
+import { connectToDatabase, SplitData, Model } from '@/service/mongo';
+import { authToken } from '@/service/utils/tools';
+
+/* 拆分数据成QA */
+export default async function handler(req: NextApiRequest, res: NextApiResponse) {
+  try {
+    const { modelId } = req.query as { modelId: string };
+    if (!modelId) {
+      throw new Error('参数错误');
+    }
+    await connectToDatabase();
+
+    const { authorization } = req.headers;
+
+    const userId = await authToken(authorization);
+
+    // 找到长度大于0的数据
+    const data = await SplitData.find({
+      userId,
+      modelId,
+      textList: { $exists: true, $not: { $size: 0 } }
+    });
+
+    jsonRes(res, {
+      data
+    });
+  } catch (err) {
+    jsonRes(res, {
+      code: 500,
+      error: err
+    });
+  }
+}
--- a/src/pages/api/model/data/splitData.ts
+++ b/src/pages/api/model/data/splitData.ts
@@ -53,7 +53,7 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse)
      textList
    });

-    // generateQA();
+    generateQA();

    jsonRes(res, {
      data: { chunks, replaceText }

--- a/src/pages/chat/components/Empty.tsx
+++ b/src/pages/chat/components/Empty.tsx
@@ -25,13 +25,13 @@ const Empty = ({ intro }: { intro: string }) => {
          <Box whiteSpace={'pre-line'}>{intro}</Box>
        </Card>
      )}
+      {/* version intro */}
      <Card p={4} mb={10}>
-        <Header>常见问题</Header>
-        <Markdown source={chatProblem} />
+        <Markdown source={versionIntro} />
      </Card>
-      {/* version intro */}
      <Card p={4}>
-        <Markdown source={versionIntro} />
+        <Header>常见问题</Header>
+        <Markdown source={chatProblem} />
      </Card>
    </Box>
  );

--- a/src/pages/model/components/ModelDataCard.tsx
+++ b/src/pages/model/components/ModelDataCard.tsx
--- a/src/pages/model/detail/components/ModelDataCard.tsx
+++ b/src/pages/model/detail/components/ModelDataCard.tsx
@@ -21,13 +21,18 @@ import {
 import type { ModelSchema } from '@/types/mongoSchema';
 import { ModelDataSchema } from '@/types/mongoSchema';
 import { ModelDataStatusMap } from '@/constants/model';
-import { usePaging } from '@/hooks/usePaging';
-import ScrollData from '@/components/ScrollData';
-import { getModelDataList, delOneModelData, putModelDataById } from '@/api/model';
+import { usePagination } from '@/hooks/usePagination';
+import {
+  getModelDataList,
+  delOneModelData,
+  putModelDataById,
+  getModelSplitDataList
+} from '@/api/model';
 import { DeleteIcon, RepeatIcon } from '@chakra-ui/icons';
 import { useToast } from '@/hooks/useToast';
 import { useLoading } from '@/hooks/useLoading';
 import dynamic from 'next/dynamic';
+import { useQuery } from '@tanstack/react-query';

 const InputModel = dynamic(() => import('./InputDataModal'));
 const SelectModel = dynamic(() => import('./SelectFileModal'));
@@ -37,16 +42,15 @@ const ModelDataCard = ({ model }: { model: ModelSchema }) => {
  const { Loading } = useLoading();

  const {
-    nextPage,
-    isLoadAll,
-    requesting,
    data: modelDataList,
+    isLoading,
+    Pagination,
    total,
-    setData,
-    getData
-  } = usePaging<ModelDataSchema>({
+    getData,
+    pageNum
+  } = usePagination<ModelDataSchema>({
    api: getModelDataList,
-    pageSize: 20,
+    pageSize: 10,
    params: {
      modelId: model._id
    }
@@ -77,6 +81,18 @@ const ModelDataCard = ({ model }: { model: ModelSchema }) => {
    onClose: onCloseSelectModal
  } = useDisclosure();

+  const { data, refetch } = useQuery(['getModelSplitDataList'], () =>
+    getModelSplitDataList(model._id)
+  );
+
+  const refetchData = useCallback(
+    (num = 1) => {
+      getData(num);
+      refetch();
+    },
+    [getData, refetch]
+  );
+
  return (
    <>
      <Flex>
@@ -91,7 +107,7 @@ const ModelDataCard = ({ model }: { model: ModelSchema }) => {
          aria-label={'refresh'}
          variant={'outline'}
          mr={4}
-          onClick={() => getData(1, true)}
+          onClick={() => refetchData(pageNum)}
        />
        <Menu>
          <MenuButton as={Button}>导入</MenuButton>
@@ -101,16 +117,9 @@ const ModelDataCard = ({ model }: { model: ModelSchema }) => {
          </MenuList>
        </Menu>
      </Flex>
-      <ScrollData
-        h={'100%'}
-        px={6}
-        mt={3}
-        isLoadAll={isLoadAll}
-        requesting={requesting}
-        nextPage={nextPage}
-        position={'relative'}
-      >
-        <TableContainer mt={4}>
+      {data && data.length > 0 && <Box fontSize={'xs'}>{data.length}条数据正在拆分中...</Box>}
+      <Box mt={4}>
+        <TableContainer h={'600px'} overflowY={'auto'}>
          <Table variant={'simple'}>
            <Thead>
              <Tr>
@@ -150,12 +159,6 @@ const ModelDataCard = ({ model }: { model: ModelSchema }) => {
                        const oldVal = modelDataList.find((data) => item._id === data._id)?.text;
                        if (oldVal !== e.target.value) {
                          updateAnswer(item._id, e.target.value);
-                          setData((state) =>
-                            state.map((data) => ({
-                              ...data,
-                              text: data._id === item._id ? e.target.value : data.text
-                            }))
-                          );
                        }
                      }}
                    ></Textarea>
@@ -170,7 +173,7 @@ const ModelDataCard = ({ model }: { model: ModelSchema }) => {
                      size={'sm'}
                      onClick={async () => {
                        delOneModelData(item._id);
-                        setData((state) => state.filter((data) => data._id !== item._id));
+                        refetchData(pageNum);
                      }}
                    />
                  </Td>
@@ -179,21 +182,17 @@ const ModelDataCard = ({ model }: { model: ModelSchema }) => {
            </Tbody>
          </Table>
        </TableContainer>
-        <Loading loading={requesting} fixed={false} />
-      </ScrollData>
+        <Box mt={2} textAlign={'end'}>
+          <Pagination />
+        </Box>
+      </Box>
+
+      <Loading loading={isLoading} fixed={false} />
      {isOpenInputModal && (
-        <InputModel
-          modelId={model._id}
-          onClose={onCloseInputModal}
-          onSuccess={() => getData(1, true)}
-        />
+        <InputModel modelId={model._id} onClose={onCloseInputModal} onSuccess={refetchData} />
      )}
      {isOpenSelectModal && (
-        <SelectModel
-          modelId={model._id}
-          onClose={onCloseSelectModal}
-          onSuccess={() => getData(1, true)}
-        />
+        <SelectModel modelId={model._id} onClose={onCloseSelectModal} onSuccess={refetchData} />
      )}
    </>
  );

--- a/src/pages/model/detail/components/SelectFileModal.tsx
+++ b/src/pages/model/detail/components/SelectFileModal.tsx
@@ -37,7 +37,7 @@ const SelectFileModal = ({
  const { File, onOpen } = useSelectFile({ fileType: fileExtension, multiple: true });
  const [fileText, setFileText] = useState('');
  const { openConfirm, ConfirmChild } = useConfirm({
-    content: '确认导入该文件，需要一定时间进行拆解，该任务无法终止！'
+    content: '确认导入该文件，需要一定时间进行拆解，该任务无法终止！如果余额不足，任务讲被终止。'
  });

  const onSelectFile = useCallback(

--- a/src/pages/model/detail/index.tsx
+++ b/src/pages/model/detail/index.tsx
@@ -11,8 +11,7 @@ import { useGlobalStore } from '@/store/global';
 import { useScreen } from '@/hooks/useScreen';
 import ModelEditForm from './components/ModelEditForm';
 import { useQuery } from '@tanstack/react-query';
-// import dynamic from 'next/dynamic';
-import ModelDataCard from './components/ModelDataCard';
+import dynamic from 'next/dynamic';

 const ModelDataCard = dynamic(() => import('./components/ModelDataCard'));

@@ -251,7 +250,6 @@ const ModelDetail = ({ modelId }: { modelId: string }) => {
        {canTrain && model._id && (
          <Card
            p={4}
-            height={'700px'}
            {...media(
              {
                gridColumnStart: 1,