feat: 模型数据管理

feat: 模型数据导入 feat: redis 向量入库 feat: 向量索引 feat: 文件导入模型 perf: 交互 perf: prompt
2023-03-29 00:22:48 +08:00
parent 713332522f
commit 2099a87908
45 changed files with 1522 additions and 284 deletions
--- a/src/pages/api/chat/chatGpt.ts
+++ b/src/pages/api/chat/chatGpt.ts
@@ -46,7 +46,7 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse)
    const model: ModelSchema = chat.modelId;
    const modelConstantsData = modelList.find((item) => item.model === model.service.modelName);
    if (!modelConstantsData) {
-      throw new Error('模型异常,请用 chatgpt 模型');
+      throw new Error('模型加载异常');
    }

    // 读取对话内容
--- a/src/pages/api/chat/vectorGpt.ts
+++ b/src/pages/api/chat/vectorGpt.ts
@@ -0,0 +1,241 @@
+import type { NextApiRequest, NextApiResponse } from 'next';
+import { createParser, ParsedEvent, ReconnectInterval } from 'eventsource-parser';
+import { connectToDatabase, ModelData } from '@/service/mongo';
+import { getOpenAIApi, authChat } from '@/service/utils/chat';
+import { httpsAgent, openaiChatFilter, systemPromptFilter } from '@/service/utils/tools';
+import { ChatCompletionRequestMessage, ChatCompletionRequestMessageRoleEnum } from 'openai';
+import { ChatItemType } from '@/types/chat';
+import { jsonRes } from '@/service/response';
+import type { ModelSchema } from '@/types/mongoSchema';
+import { PassThrough } from 'stream';
+import { modelList } from '@/constants/model';
+import { pushChatBill } from '@/service/events/pushBill';
+import { connectRedis } from '@/service/redis';
+import { VecModelDataIndex } from '@/constants/redis';
+import { vectorToBuffer } from '@/utils/tools';
+
+let vectorData = [
+  -0.025028639, -0.010407282, 0.026523087, -0.0107438695, -0.006967359, 0.010043768, -0.012043097,
+  0.008724345, -0.028919589, -0.0117738275, 0.0050690062, 0.02961969
+].concat(new Array(1524).fill(0));
+
+/* 发送提示词 */
+export default async function handler(req: NextApiRequest, res: NextApiResponse) {
+  let step = 0; // step=1时，表示开始了流响应
+  const stream = new PassThrough();
+  stream.on('error', () => {
+    console.log('error: ', 'stream error');
+    stream.destroy();
+  });
+  res.on('close', () => {
+    stream.destroy();
+  });
+  res.on('error', () => {
+    console.log('error: ', 'request error');
+    stream.destroy();
+  });
+
+  try {
+    const { chatId, prompt } = req.body as {
+      prompt: ChatItemType;
+      chatId: string;
+    };
+
+    const { authorization } = req.headers;
+    if (!chatId || !prompt) {
+      throw new Error('缺少参数');
+    }
+
+    await connectToDatabase();
+    const redis = await connectRedis();
+
+    const { chat, userApiKey, systemKey, userId } = await authChat(chatId, authorization);
+
+    const model: ModelSchema = chat.modelId;
+    const modelConstantsData = modelList.find((item) => item.model === model.service.modelName);
+    if (!modelConstantsData) {
+      throw new Error('模型加载异常');
+    }
+
+    // 读取对话内容
+    const prompts = [...chat.content, prompt];
+
+    // 获取 chatAPI
+    const chatAPI = getOpenAIApi(userApiKey || systemKey);
+
+    // 把输入的内容转成向量
+    const promptVector = await chatAPI
+      .createEmbedding(
+        {
+          model: 'text-embedding-ada-002',
+          input: prompt.value
+        },
+        {
+          timeout: 120000,
+          httpsAgent
+        }
+      )
+      .then((res) => res?.data?.data?.[0]?.embedding || []);
+
+    const binary = vectorToBuffer(promptVector);
+
+    // 搜索系统提示词, 按相似度从 redis 中搜出前3条不同 dataId 的数据
+    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)}}=>[KNN 10 @vector $blob AS score]`,
+      'RETURN',
+      '1',
+      'dataId',
+      // 'SORTBY',
+      // 'score',
+      'PARAMS',
+      '2',
+      'blob',
+      binary,
+      'DIALECT',
+      '2'
+    ]);
+
+    // 格式化响应值，获取去重后的id
+    let formatIds = [2, 4, 6, 8, 10, 12, 14, 16, 18, 20]
+      .map((i) => {
+        if (!redisData[i] || !redisData[i][1]) return '';
+        return redisData[i][1];
+      })
+      .filter((item) => item);
+    formatIds = Array.from(new Set(formatIds));
+
+    if (formatIds.length === 0) {
+      throw new Error('对不起，我没有找到你的问题');
+    }
+
+    // 从 mongo 中取出原文作为提示词
+    const textArr = (
+      await Promise.all(
+        [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 || '');
+        })
+      )
+    ).filter((item) => item);
+
+    // textArr 筛选，最多 3000 tokens
+    const systemPrompt = systemPromptFilter(textArr, 2800);
+
+    prompts.unshift({
+      obj: 'SYSTEM',
+      value: `请根据下面的知识回答问题： ${systemPrompt}`
+    });
+
+    // 控制在 tokens 数量，防止超出
+    const filterPrompts = openaiChatFilter(prompts, modelConstantsData.contextMaxToken);
+
+    // 格式化文本内容成 chatgpt 格式
+    const map = {
+      Human: ChatCompletionRequestMessageRoleEnum.User,
+      AI: ChatCompletionRequestMessageRoleEnum.Assistant,
+      SYSTEM: ChatCompletionRequestMessageRoleEnum.System
+    };
+    const formatPrompts: ChatCompletionRequestMessage[] = filterPrompts.map(
+      (item: ChatItemType) => ({
+        role: map[item.obj],
+        content: item.value
+      })
+    );
+    // console.log(formatPrompts);
+    // 计算温度
+    const temperature = modelConstantsData.maxTemperature * (model.temperature / 10);
+
+    let startTime = Date.now();
+    // 发出请求
+    const chatResponse = await chatAPI.createChatCompletion(
+      {
+        model: model.service.chatModel,
+        temperature: temperature,
+        // max_tokens: modelConstantsData.maxToken,
+        messages: formatPrompts,
+        frequency_penalty: 0.5, // 越大，重复内容越少
+        presence_penalty: -0.5, // 越大，越容易出现新内容
+        stream: true
+      },
+      {
+        timeout: 40000,
+        responseType: 'stream',
+        httpsAgent
+      }
+    );
+
+    console.log('api response time:', `${(Date.now() - startTime) / 1000}s`);
+
+    // 创建响应流
+    res.setHeader('Content-Type', 'text/event-stream;charset-utf-8');
+    res.setHeader('Access-Control-Allow-Origin', '*');
+    res.setHeader('X-Accel-Buffering', 'no');
+    res.setHeader('Cache-Control', 'no-cache, no-transform');
+    step = 1;
+
+    let responseContent = '';
+    stream.pipe(res);
+
+    const onParse = async (event: ParsedEvent | ReconnectInterval) => {
+      if (event.type !== 'event') return;
+      const data = event.data;
+      if (data === '[DONE]') return;
+      try {
+        const json = JSON.parse(data);
+        const content: string = json?.choices?.[0].delta.content || '';
+        if (!content || (responseContent === '' && content === '\n')) return;
+
+        responseContent += content;
+        // console.log('content:', content)
+        !stream.destroyed && stream.push(content.replace(/\n/g, '<br/>'));
+      } catch (error) {
+        error;
+      }
+    };
+
+    const decoder = new TextDecoder();
+    try {
+      for await (const chunk of chatResponse.data as any) {
+        if (stream.destroyed) {
+          // 流被中断了，直接忽略后面的内容
+          break;
+        }
+        const parser = createParser(onParse);
+        parser.feed(decoder.decode(chunk));
+      }
+    } catch (error) {
+      console.log('pipe error', error);
+    }
+    // close stream
+    !stream.destroyed && stream.push(null);
+    stream.destroy();
+
+    const promptsContent = formatPrompts.map((item) => item.content).join('');
+    // 只有使用平台的 key 才计费
+    pushChatBill({
+      isPay: !userApiKey,
+      modelName: model.service.modelName,
+      userId,
+      chatId,
+      text: promptsContent + responseContent
+    });
+    // jsonRes(res);
+  } catch (err: any) {
+    if (step === 1) {
+      // 直接结束流
+      console.log('error，结束');
+      stream.destroy();
+    } else {
+      res.status(500);
+      jsonRes(res, {
+        code: 500,
+        error: err
+      });
+    }
+  }
+}
--- a/src/pages/api/data/splitData.ts
+++ b/src/pages/api/data/splitData.ts
@@ -24,7 +24,7 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse)
    if (!DataRecord) {
      throw new Error('找不到数据集');
    }
-    const replaceText = text.replace(/[\r\n\\n]+/g, ' ');
+    const replaceText = text.replace(/[\\n]+/g, ' ');

    // 文本拆分成 chunk
    let chunks = replaceText.match(/[^!?.。]+[!?.。]/g) || [];
@@ -35,7 +35,7 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse)
    chunks.forEach((chunk) => {
      splitText += chunk;
      const tokens = encode(splitText).length;
-      if (tokens >= 980) {
+      if (tokens >= 780) {
        dataItems.push({
          userId,
          dataId,
--- a/src/pages/api/model/create.ts
+++ b/src/pages/api/model/create.ts
@@ -3,7 +3,7 @@ import type { NextApiRequest, NextApiResponse } from 'next';
 import { jsonRes } from '@/service/response';
 import { connectToDatabase } from '@/service/mongo';
 import { authToken } from '@/service/utils/tools';
-import { ModelStatusEnum, modelList, ChatModelNameEnum } from '@/constants/model';
+import { ModelStatusEnum, modelList, ChatModelNameEnum, ChatModelNameMap } from '@/constants/model';
 import { Model } from '@/service/models/model';

 export default async function handler(req: NextApiRequest, res: NextApiResponse<any>) {
@@ -33,15 +33,6 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse<

    await connectToDatabase();

-    // 重名校验
-    const authRepeatName = await Model.findOne({
-      name,
-      userId
-    });
-    if (authRepeatName) {
-      throw new Error('模型名重复');
-    }
-
    // 上限校验
    const authCount = await Model.countDocuments({
      userId
@@ -57,9 +48,9 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse<
      status: ModelStatusEnum.running,
      service: {
        company: modelItem.serviceCompany,
-        trainId: modelItem.trainName,
-        chatModel: modelItem.model,
-        modelName: modelItem.model
+        trainId: '',
+        chatModel: ChatModelNameMap[modelItem.model], // 聊天时用的模型
+        modelName: modelItem.model // 最底层的模型，不会变，用于计费等核心操作
      }
    });

--- a/src/pages/api/model/data/delModelDataById.ts
+++ b/src/pages/api/model/data/delModelDataById.ts
@@ -5,8 +5,8 @@ import { authToken } from '@/service/utils/tools';

 export default async function handler(req: NextApiRequest, res: NextApiResponse<any>) {
  try {
-    let { modelId } = req.query as {
-      modelId: string;
+    let { dataId } = req.query as {
+      dataId: string;
    };
    const { authorization } = req.headers;

@@ -14,7 +14,7 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse<
      throw new Error('无权操作');
    }

-    if (!modelId) {
+    if (!dataId) {
      throw new Error('缺少参数');
    }

@@ -24,7 +24,7 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse<
    await connectToDatabase();

    await ModelData.deleteOne({
-      modelId,
+      _id: dataId,
      userId
    });

--- a/src/pages/api/model/data/getModelData.ts
+++ b/src/pages/api/model/data/getModelData.ts
@@ -14,6 +14,7 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse<
      pageNum: string;
      pageSize: string;
    };
+
    const { authorization } = req.headers;

    pageNum = +pageNum;
@@ -41,7 +42,15 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse<
      .limit(pageSize);

    jsonRes(res, {
-      data
+      data: {
+        pageNum,
+        pageSize,
+        data,
+        total: await ModelData.countDocuments({
+          modelId,
+          userId
+        })
+      }
    });
  } catch (err) {
    jsonRes(res, {
--- a/src/pages/api/model/data/pushModelDataInput.ts
+++ b/src/pages/api/model/data/pushModelDataInput.ts
@@ -2,12 +2,14 @@ import type { NextApiRequest, NextApiResponse } from 'next';
 import { jsonRes } from '@/service/response';
 import { connectToDatabase, ModelData, Model } from '@/service/mongo';
 import { authToken } from '@/service/utils/tools';
+import { ModelDataSchema } from '@/types/mongoSchema';
+import { generateVector } from '@/service/events/generateVector';

 export default async function handler(req: NextApiRequest, res: NextApiResponse<any>) {
  try {
    const { modelId, data } = req.body as {
      modelId: string;
-      data: { q: string; a: string }[];
+      data: { text: ModelDataSchema['text']; q: ModelDataSchema['q'] }[];
    };
    const { authorization } = req.headers;

@@ -43,6 +45,8 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse<
      }))
    );

+    generateVector(true);
+
    jsonRes(res, {
      data: model
    });
--- a/src/pages/api/model/data/pushModelDataSelectData.ts
+++ b/src/pages/api/model/data/pushModelDataSelectData.ts
@@ -0,0 +1,57 @@
+import type { NextApiRequest, NextApiResponse } from 'next';
+import { jsonRes } from '@/service/response';
+import { connectToDatabase, DataItem, ModelData } from '@/service/mongo';
+import { authToken } from '@/service/utils/tools';
+import { customAlphabet } from 'nanoid';
+const nanoid = customAlphabet('abcdefghijklmnopqrstuvwxyz1234567890', 12);
+
+export default async function handler(req: NextApiRequest, res: NextApiResponse) {
+  try {
+    let { dataIds, modelId } = req.body as { dataIds: string[]; modelId: string };
+
+    if (!dataIds) {
+      throw new Error('参数错误');
+    }
+    await connectToDatabase();
+
+    const { authorization } = req.headers;
+
+    const userId = await authToken(authorization);
+
+    const dataItems = (
+      await Promise.all(
+        dataIds.map((dataId) =>
+          DataItem.find<{ _id: string; result: { q: string }[]; text: string }>(
+            {
+              userId,
+              dataId
+            },
+            'result text'
+          )
+        )
+      )
+    ).flat();
+
+    // push data
+    await ModelData.insertMany(
+      dataItems.map((item) => ({
+        modelId: modelId,
+        userId,
+        text: item.text,
+        q: item.result.map((item) => ({
+          id: nanoid(),
+          text: item.q
+        }))
+      }))
+    );
+
+    jsonRes(res, {
+      data: dataItems
+    });
+  } catch (err) {
+    jsonRes(res, {
+      code: 500,
+      error: err
+    });
+  }
+}
--- a/src/pages/api/model/data/putModelData.ts
+++ b/src/pages/api/model/data/putModelData.ts
@@ -5,9 +5,9 @@ import { authToken } from '@/service/utils/tools';

 export default async function handler(req: NextApiRequest, res: NextApiResponse<any>) {
  try {
-    let { modelId, answer } = req.body as {
-      modelId: string;
-      answer: string;
+    let { dataId, text } = req.body as {
+      dataId: string;
+      text: string;
    };
    const { authorization } = req.headers;

@@ -15,7 +15,7 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse<
      throw new Error('无权操作');
    }

-    if (!modelId) {
+    if (!dataId) {
      throw new Error('缺少参数');
    }

@@ -26,11 +26,11 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse<

    await ModelData.updateOne(
      {
-        modelId,
+        _id: dataId,
        userId
      },
      {
-        a: answer
+        text
      }
    );

--- a/src/pages/api/model/data/splitData.ts
+++ b/src/pages/api/model/data/splitData.ts
@@ -0,0 +1,67 @@
+import type { NextApiRequest, NextApiResponse } from 'next';
+import { jsonRes } from '@/service/response';
+import { connectToDatabase, SplitData, Model } from '@/service/mongo';
+import { authToken } from '@/service/utils/tools';
+import { generateQA } from '@/service/events/generateQA';
+import { encode } from 'gpt-token-utils';
+
+/* 拆分数据成QA */
+export default async function handler(req: NextApiRequest, res: NextApiResponse) {
+  try {
+    const { text, modelId } = req.body as { text: string; modelId: string };
+    if (!text || !modelId) {
+      throw new Error('参数错误');
+    }
+    await connectToDatabase();
+
+    const { authorization } = req.headers;
+
+    const userId = await authToken(authorization);
+
+    // 验证是否是该用户的 model
+    const model = await Model.findOne({
+      _id: modelId,
+      userId
+    });
+
+    if (!model) {
+      throw new Error('无权操作该模型');
+    }
+
+    const replaceText = text.replace(/(\\n|\n)+/g, ' ');
+
+    // 文本拆分成 chunk
+    let chunks = replaceText.match(/[^!?.。]+[!?.。]/g) || [];
+
+    const textList: string[] = [];
+    let splitText = '';
+
+    chunks.forEach((chunk) => {
+      splitText += chunk;
+      const tokens = encode(splitText).length;
+      if (tokens >= 980) {
+        textList.push(splitText);
+        splitText = '';
+      }
+    });
+
+    // 批量插入数据
+    await SplitData.create({
+      userId,
+      modelId,
+      rawText: text,
+      textList
+    });
+
+    // generateQA();
+
+    jsonRes(res, {
+      data: { chunks, replaceText }
+    });
+  } catch (err) {
+    jsonRes(res, {
+      code: 500,
+      error: err
+    });
+  }
+}
--- a/src/pages/api/model/del.ts
+++ b/src/pages/api/model/del.ts
@@ -1,6 +1,6 @@
 import type { NextApiRequest, NextApiResponse } from 'next';
 import { jsonRes } from '@/service/response';
-import { Chat, Model, Training, connectToDatabase } from '@/service/mongo';
+import { Chat, Model, Training, connectToDatabase, ModelData } from '@/service/mongo';
 import { authToken, getUserOpenaiKey } from '@/service/utils/tools';
 import { TrainingStatusEnum } from '@/constants/model';
 import { getOpenAIApi } from '@/service/utils/chat';
@@ -26,16 +26,20 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse<

    await connectToDatabase();

-    // 删除模型
-    await Model.deleteOne({
-      _id: modelId,
-      userId
-    });
-
+    let requestQueue: any[] = [];
    // 删除对应的聊天
-    await Chat.deleteMany({
-      modelId
-    });
+    requestQueue.push(
+      Chat.deleteMany({
+        modelId
+      })
+    );
+
+    // 删除数据集
+    requestQueue.push(
+      ModelData.deleteMany({
+        modelId
+      })
+    );

    // 查看是否正在训练
    const training: TrainingItemType | null = await Training.findOne({
@@ -56,9 +60,20 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse<
    }

    // 删除对应训练记录
-    await Training.deleteMany({
-      modelId
-    });
+    requestQueue.push(
+      Training.deleteMany({
+        modelId
+      })
+    );
+
+    // 删除模型
+    requestQueue.push(
+      Model.deleteOne({
+        _id: modelId,
+        userId
+      })
+    );
+    await requestQueue;

    jsonRes(res);
  } catch (err) {
--- a/src/pages/api/model/update.ts
+++ b/src/pages/api/model/update.ts
@@ -37,7 +37,7 @@ export default async function handler(req: NextApiRequest, res: NextApiResponse<
        systemPrompt,
        intro,
        temperature,
-        service,
+        // service,
        security
      }
    );
--- a/src/pages/chat/index.tsx
+++ b/src/pages/chat/index.tsx
@@ -119,6 +119,7 @@ const Chat = ({ chatId }: { chatId: string }) => {
    async (prompts: ChatSiteItemType) => {
      const urlMap: Record<string, string> = {
        [ChatModelNameEnum.GPT35]: '/api/chat/chatGpt',
+        [ChatModelNameEnum.VECTOR_GPT]: '/api/chat/vectorGpt',
        [ChatModelNameEnum.GPT3]: '/api/chat/gpt3'
      };

--- a/src/pages/data/list.tsx
+++ b/src/pages/data/list.tsx
@@ -184,7 +184,7 @@ const DataList = () => {
                  >
                    导入
                  </Button>
-                  <Menu>
+                  {/* <Menu>
                    <MenuButton as={Button} mr={2} size={'sm'} isLoading={isExporting}>
                      导出
                    </MenuButton>
@@ -200,7 +200,7 @@ const DataList = () => {
                        </MenuItem>
                      )}
                    </MenuList>
-                  </Menu>
+                  </Menu> */}

                  <Button
                    size={'sm'}
--- a/src/pages/model/detail/components/InputDataModal.tsx
+++ b/src/pages/model/detail/components/InputDataModal.tsx
@@ -0,0 +1,141 @@
+import React, { useState, useCallback } from 'react';
+import {
+  Box,
+  IconButton,
+  Flex,
+  Button,
+  Modal,
+  ModalOverlay,
+  ModalContent,
+  ModalHeader,
+  ModalCloseButton,
+  Input,
+  Textarea
+} from '@chakra-ui/react';
+import { useForm, useFieldArray } from 'react-hook-form';
+import { postModelDataInput } from '@/api/model';
+import { useToast } from '@/hooks/useToast';
+import { DeleteIcon } from '@chakra-ui/icons';
+import { customAlphabet } from 'nanoid';
+const nanoid = customAlphabet('abcdefghijklmnopqrstuvwxyz1234567890', 12);
+
+type FormData = { text: string; q: { val: string }[] };
+
+const InputDataModal = ({
+  onClose,
+  onSuccess,
+  modelId
+}: {
+  onClose: () => void;
+  onSuccess: () => void;
+  modelId: string;
+}) => {
+  const [importing, setImporting] = useState(false);
+  const { toast } = useToast();
+
+  const { register, handleSubmit, control } = useForm<FormData>({
+    defaultValues: {
+      text: '',
+      q: [{ val: '' }]
+    }
+  });
+  const {
+    fields: inputQ,
+    append: appendQ,
+    remove: removeQ
+  } = useFieldArray({
+    control,
+    name: 'q'
+  });
+
+  const sureImportData = useCallback(
+    async (e: FormData) => {
+      setImporting(true);
+
+      try {
+        await postModelDataInput({
+          modelId: modelId,
+          data: [
+            {
+              text: e.text,
+              q: e.q.map((item) => ({
+                id: nanoid(),
+                text: item.val
+              }))
+            }
+          ]
+        });
+
+        toast({
+          title: '导入数据成功,需要一段时间训练',
+          status: 'success'
+        });
+        onClose();
+        onSuccess();
+      } catch (err) {
+        console.log(err);
+      }
+      setImporting(false);
+    },
+    [modelId, onClose, onSuccess, toast]
+  );
+
+  return (
+    <Modal isOpen={true} onClose={onClose}>
+      <ModalOverlay />
+      <ModalContent maxW={'min(900px, 90vw)'} maxH={'80vh'} position={'relative'}>
+        <ModalHeader>手动导入</ModalHeader>
+        <ModalCloseButton />
+        <Box px={6} pb={2} overflowY={'auto'}>
+          <Box mb={2}>知识点:</Box>
+          <Textarea
+            mb={4}
+            placeholder="知识点"
+            rows={3}
+            maxH={'200px'}
+            {...register(`text`, {
+              required: '知识点'
+            })}
+          />
+          {inputQ.map((item, index) => (
+            <Box key={item.id} mb={5}>
+              <Box mb={2}>问法{index + 1}:</Box>
+              <Flex>
+                <Input
+                  placeholder="问法"
+                  {...register(`q.${index}.val`, {
+                    required: '问法不能为空'
+                  })}
+                ></Input>
+                {inputQ.length > 1 && (
+                  <IconButton
+                    icon={<DeleteIcon />}
+                    aria-label={'delete'}
+                    colorScheme={'gray'}
+                    variant={'unstyled'}
+                    onClick={() => removeQ(index)}
+                  />
+                )}
+              </Flex>
+            </Box>
+          ))}
+        </Box>
+
+        <Flex px={6} pt={2} pb={4}>
+          <Button alignSelf={'flex-start'} variant={'outline'} onClick={() => appendQ({ val: '' })}>
+            增加问法
+          </Button>
+          <Box flex={1}></Box>
+          <Button variant={'outline'} mr={3} onClick={onClose}>
+            取消
+          </Button>
+          <Button isLoading={importing} onClick={handleSubmit(sureImportData)}>
+            确认导入
+          </Button>
+        </Flex>
+      </ModalContent>
+    </Modal>
+  );
+};
+
+export default InputDataModal;
--- a/src/pages/model/detail/components/ModelDataCard.tsx
+++ b/src/pages/model/detail/components/ModelDataCard.tsx
@@ -0,0 +1,202 @@
+import React, { useCallback } from 'react';
+import {
+  Box,
+  TableContainer,
+  Table,
+  Thead,
+  Tbody,
+  Tr,
+  Th,
+  Td,
+  IconButton,
+  Flex,
+  Button,
+  useDisclosure,
+  Textarea,
+  Menu,
+  MenuButton,
+  MenuList,
+  MenuItem
+} from '@chakra-ui/react';
+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 { DeleteIcon, RepeatIcon } from '@chakra-ui/icons';
+import { useToast } from '@/hooks/useToast';
+import { useLoading } from '@/hooks/useLoading';
+import dynamic from 'next/dynamic';
+
+const InputModel = dynamic(() => import('./InputDataModal'));
+const SelectModel = dynamic(() => import('./SelectFileModal'));
+
+const ModelDataCard = ({ model }: { model: ModelSchema }) => {
+  const { toast } = useToast();
+  const { Loading } = useLoading();
+
+  const {
+    nextPage,
+    isLoadAll,
+    requesting,
+    data: modelDataList,
+    total,
+    setData,
+    getData
+  } = usePaging<ModelDataSchema>({
+    api: getModelDataList,
+    pageSize: 20,
+    params: {
+      modelId: model._id
+    }
+  });
+
+  const updateAnswer = useCallback(
+    async (dataId: string, text: string) => {
+      await putModelDataById({
+        dataId,
+        text
+      });
+      toast({
+        title: '修改回答成功',
+        status: 'success'
+      });
+    },
+    [toast]
+  );
+
+  const {
+    isOpen: isOpenInputModal,
+    onOpen: onOpenInputModal,
+    onClose: onCloseInputModal
+  } = useDisclosure();
+  const {
+    isOpen: isOpenSelectModal,
+    onOpen: onOpenSelectModal,
+    onClose: onCloseSelectModal
+  } = useDisclosure();
+
+  return (
+    <>
+      <Flex>
+        <Box fontWeight={'bold'} fontSize={'lg'} flex={1}>
+          模型数据: {total}组{' '}
+          <Box as={'span'} fontSize={'sm'}>
+            （测试版本）
+          </Box>
+        </Box>
+        <IconButton
+          icon={<RepeatIcon />}
+          aria-label={'refresh'}
+          variant={'outline'}
+          mr={4}
+          onClick={() => getData(1, true)}
+        />
+        <Menu>
+          <MenuButton as={Button}>导入</MenuButton>
+          <MenuList>
+            <MenuItem onClick={onOpenInputModal}>手动输入</MenuItem>
+            <MenuItem onClick={onOpenSelectModal}>文件导入</MenuItem>
+          </MenuList>
+        </Menu>
+      </Flex>
+      <ScrollData
+        h={'100%'}
+        px={6}
+        mt={3}
+        isLoadAll={isLoadAll}
+        requesting={requesting}
+        nextPage={nextPage}
+        position={'relative'}
+      >
+        <TableContainer mt={4}>
+          <Table variant={'simple'}>
+            <Thead>
+              <Tr>
+                <Th>Question</Th>
+                <Th>Text</Th>
+                <Th>Status</Th>
+                <Th></Th>
+              </Tr>
+            </Thead>
+            <Tbody>
+              {modelDataList.map((item) => (
+                <Tr key={item._id}>
+                  <Td w={'350px'}>
+                    {item.q.map((item, i) => (
+                      <Box
+                        key={item.id}
+                        fontSize={'xs'}
+                        w={'100%'}
+                        whiteSpace={'pre-wrap'}
+                        _notLast={{ mb: 1 }}
+                      >
+                        Q{i + 1}:{' '}
+                        <Box as={'span'} userSelect={'all'}>
+                          {item.text}
+                        </Box>
+                      </Box>
+                    ))}
+                  </Td>
+                  <Td minW={'200px'}>
+                    <Textarea
+                      w={'100%'}
+                      h={'100%'}
+                      defaultValue={item.text}
+                      fontSize={'xs'}
+                      resize={'both'}
+                      onBlur={(e) => {
+                        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>
+                  </Td>
+                  <Td w={'100px'}>{ModelDataStatusMap[item.status]}</Td>
+                  <Td>
+                    <IconButton
+                      icon={<DeleteIcon />}
+                      variant={'outline'}
+                      colorScheme={'gray'}
+                      aria-label={'delete'}
+                      size={'sm'}
+                      onClick={async () => {
+                        delOneModelData(item._id);
+                        setData((state) => state.filter((data) => data._id !== item._id));
+                      }}
+                    />
+                  </Td>
+                </Tr>
+              ))}
+            </Tbody>
+          </Table>
+        </TableContainer>
+        <Loading loading={requesting} fixed={false} />
+      </ScrollData>
+      {isOpenInputModal && (
+        <InputModel
+          modelId={model._id}
+          onClose={onCloseInputModal}
+          onSuccess={() => getData(1, true)}
+        />
+      )}
+      {isOpenSelectModal && (
+        <SelectModel
+          modelId={model._id}
+          onClose={onCloseSelectModal}
+          onSuccess={() => getData(1, true)}
+        />
+      )}
+    </>
+  );
+};
+
+export default ModelDataCard;
--- a/src/pages/model/detail/components/ModelEditForm.tsx
+++ b/src/pages/model/detail/components/ModelEditForm.tsx
@@ -11,13 +11,28 @@ import {
  SliderFilledTrack,
  SliderThumb,
  SliderMark,
-  Tooltip
+  Tooltip,
+  Button
 } from '@chakra-ui/react';
 import { QuestionOutlineIcon } from '@chakra-ui/icons';
 import type { ModelSchema } from '@/types/mongoSchema';
 import { UseFormReturn } from 'react-hook-form';
+import { modelList } from '@/constants/model';
+import { formatPrice } from '@/utils/user';
+import { useConfirm } from '@/hooks/useConfirm';

-const ModelEditForm = ({ formHooks }: { formHooks: UseFormReturn<ModelSchema> }) => {
+const ModelEditForm = ({
+  formHooks,
+  canTrain,
+  handleDelModel
+}: {
+  formHooks: UseFormReturn<ModelSchema>;
+  canTrain: boolean;
+  handleDelModel: () => void;
+}) => {
+  const { openConfirm, ConfirmChild } = useConfirm({
+    content: '确认删除该模型?'
+  });
  const { register, setValue, getValues } = formHooks;
  const [refresh, setRefresh] = useState(false);

@@ -29,7 +44,7 @@ const ModelEditForm = ({ formHooks }: { formHooks: UseFormReturn<ModelSchema> })
        </Flex>
        <FormControl mt={4}>
          <Flex alignItems={'center'}>
-            <Box flex={'0 0 50px'} w={0}>
+            <Box flex={'0 0 80px'} w={0}>
              名称:
            </Box>
            <Input
@@ -39,7 +54,36 @@ const ModelEditForm = ({ formHooks }: { formHooks: UseFormReturn<ModelSchema> })
            ></Input>
          </Flex>
        </FormControl>
-        <FormControl mt={4}>
+        <Flex alignItems={'center'} mt={4}>
+          <Box flex={'0 0 80px'} w={0}>
+            底层模型:
+          </Box>
+          <Box>{getValues('service.modelName')}</Box>
+        </Flex>
+        <Flex alignItems={'center'} mt={4}>
+          <Box flex={'0 0 80px'} w={0}>
+            价格:
+          </Box>
+          <Box>
+            {formatPrice(
+              modelList.find((item) => item.model === getValues('service.modelName'))?.price || 0,
+              1000
+            )}
+            元/1K tokens(包括上下文和回答)
+          </Box>
+        </Flex>
+        <Flex mt={5} alignItems={'center'}>
+          <Box flex={'0 0 80px'}>删除:</Box>
+          <Button
+            colorScheme={'gray'}
+            variant={'outline'}
+            size={'sm'}
+            onClick={openConfirm(handleDelModel)}
+          >
+            删除模型
+          </Button>
+        </Flex>
+        {/* <FormControl mt={4}>
          <Box mb={1}>介绍:</Box>
          <Textarea
            rows={5}
@@ -47,7 +91,7 @@ const ModelEditForm = ({ formHooks }: { formHooks: UseFormReturn<ModelSchema> })
            {...register('intro')}
            placeholder={'模型的介绍，仅做展示，不影响模型的效果'}
          />
-        </FormControl>
+        </FormControl> */}
      </Card>
      <Card p={4}>
        <Box fontWeight={'bold'}>模型效果</Box>
@@ -94,15 +138,24 @@ const ModelEditForm = ({ formHooks }: { formHooks: UseFormReturn<ModelSchema> })
          </Flex>
        </FormControl>
        <Box mt={4}>
-          <Box mb={1}>系统提示词</Box>
-          <Textarea
-            rows={6}
-            maxLength={-1}
-            {...register('systemPrompt')}
-            placeholder={
-              '模型默认的 prompt 词，通过调整该内容，可以生成一个限定范围的模型。\n\n注意，改功能会影响对话的整体朝向！'
-            }
-          />
+          {canTrain ? (
+            <Box fontWeight={'bold'}>
+              训练的模型会自动根据知识库内容回答，无法设置系统prompt。注意：
+              使用该模型，在对话时需要消耗等多的 tokens
+            </Box>
+          ) : (
+            <>
+              <Box mb={1}>系统提示词</Box>
+              <Textarea
+                rows={6}
+                maxLength={-1}
+                {...register('systemPrompt')}
+                placeholder={
+                  '模型默认的 prompt 词，通过调整该内容，可以生成一个限定范围的模型。\n\n注意，改功能会影响对话的整体朝向！'
+                }
+              />
+            </>
+          )}
        </Box>
      </Card>
      {/* <Card p={4}>
@@ -202,6 +255,7 @@ const ModelEditForm = ({ formHooks }: { formHooks: UseFormReturn<ModelSchema> })
          </Flex>
        </FormControl>
      </Card> */}
+      <ConfirmChild />
    </>
  );
 };
--- a/src/pages/model/detail/components/SelectFileModal.tsx
+++ b/src/pages/model/detail/components/SelectFileModal.tsx
@@ -0,0 +1,155 @@
+import React, { useState, useCallback } from 'react';
+import {
+  Box,
+  Flex,
+  Button,
+  Modal,
+  ModalOverlay,
+  ModalContent,
+  ModalHeader,
+  ModalCloseButton,
+  ModalBody
+} from '@chakra-ui/react';
+import { useToast } from '@/hooks/useToast';
+import { useSelectFile } from '@/hooks/useSelectFile';
+import { customAlphabet } from 'nanoid';
+import { encode } from 'gpt-token-utils';
+import { useConfirm } from '@/hooks/useConfirm';
+import { readTxtContent, readPdfContent, readDocContent } from '@/utils/tools';
+import { useMutation } from '@tanstack/react-query';
+import { postModelDataFileText } from '@/api/model';
+
+const nanoid = customAlphabet('abcdefghijklmnopqrstuvwxyz1234567890', 12);
+
+const fileExtension = '.txt,.doc,.docx,.pdf,.md';
+
+const SelectFileModal = ({
+  onClose,
+  onSuccess,
+  modelId
+}: {
+  onClose: () => void;
+  onSuccess: () => void;
+  modelId: string;
+}) => {
+  const [selecting, setSelecting] = useState(false);
+  const { toast } = useToast();
+  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！'
+  });
+
+  const onSelectFile = useCallback(
+    async (e: File[]) => {
+      setSelecting(true);
+      try {
+        const fileTexts = (
+          await Promise.all(
+            e.map((file) => {
+              // @ts-ignore
+              const extension = file?.name?.split('.').pop().toLowerCase();
+              switch (extension) {
+                case 'txt':
+                case 'md':
+                  return readTxtContent(file);
+                case 'pdf':
+                  return readPdfContent(file);
+                case 'doc':
+                case 'docx':
+                  return readDocContent(file);
+                default:
+                  return '';
+              }
+            })
+          )
+        )
+          .join('\n')
+          .replace(/\n+/g, '\n');
+        setFileText(fileTexts);
+        console.log(encode(fileTexts));
+      } catch (error: any) {
+        console.log(error);
+        toast({
+          title: typeof error === 'string' ? error : '解析文件失败',
+          status: 'error'
+        });
+      }
+      setSelecting(false);
+    },
+    [setSelecting, toast]
+  );
+
+  const { mutate, isLoading } = useMutation({
+    mutationFn: async () => {
+      if (!fileText) return;
+      await postModelDataFileText(modelId, fileText);
+      toast({
+        title: '导入数据成功,需要一段拆解和训练',
+        status: 'success'
+      });
+      onClose();
+      onSuccess();
+    },
+    onError() {
+      toast({
+        title: '导入文件失败',
+        status: 'error'
+      });
+    }
+  });
+
+  return (
+    <Modal isOpen={true} onClose={onClose}>
+      <ModalOverlay />
+      <ModalContent maxW={'min(900px, 90vw)'} position={'relative'}>
+        <ModalHeader>文件导入</ModalHeader>
+        <ModalCloseButton />
+
+        <ModalBody>
+          <Flex
+            flexDirection={'column'}
+            p={2}
+            h={'100%'}
+            alignItems={'center'}
+            justifyContent={'center'}
+            fontSize={'sm'}
+          >
+            <Button isLoading={selecting} onClick={onOpen}>
+              选择文件
+            </Button>
+            <Box mt={2}>支持 {fileExtension} 文件. 会先对文本进行拆分，需要时间较长。</Box>
+            <Box mt={2}>
+              一共 {fileText.length} 个字，{encode(fileText).length} 个tokens
+            </Box>
+            <Box
+              h={'300px'}
+              w={'100%'}
+              overflow={'auto'}
+              p={2}
+              backgroundColor={'blackAlpha.50'}
+              whiteSpace={'pre'}
+              fontSize={'xs'}
+            >
+              {fileText}
+            </Box>
+          </Flex>
+        </ModalBody>
+
+        <Flex px={6} pt={2} pb={4}>
+          <Box flex={1}></Box>
+          <Button variant={'outline'} mr={3} onClick={onClose}>
+            取消
+          </Button>
+          <Button isLoading={isLoading} isDisabled={fileText === ''} onClick={openConfirm(mutate)}>
+            确认导入
+          </Button>
+        </Flex>
+      </ModalContent>
+      <ConfirmChild />
+      <File onSelect={onSelectFile} />
+    </Modal>
+  );
+};
+
+export default SelectFileModal;
--- a/src/pages/model/detail/components/Training.tsx
+++ b/src/pages/model/detail/components/Training.tsx
--- a/src/pages/model/detail/index.tsx
+++ b/src/pages/model/detail/index.tsx
@@ -1,37 +1,27 @@
 import React, { useCallback, useState, useRef, useMemo, useEffect } from 'react';
 import { useRouter } from 'next/router';
-import {
-  getModelById,
-  delModelById,
-  postTrainModel,
-  putModelTrainingStatus,
-  putModelById
-} from '@/api/model';
+import { getModelById, delModelById, putModelTrainingStatus, putModelById } from '@/api/model';
 import { getChatSiteId } from '@/api/chat';
 import type { ModelSchema } from '@/types/mongoSchema';
 import { Card, Box, Flex, Button, Tag, Grid } from '@chakra-ui/react';
 import { useToast } from '@/hooks/useToast';
-import { useConfirm } from '@/hooks/useConfirm';
 import { useForm } from 'react-hook-form';
 import { formatModelStatus, ModelStatusEnum, modelList, defaultModel } from '@/constants/model';
 import { useGlobalStore } from '@/store/global';
 import { useScreen } from '@/hooks/useScreen';
 import ModelEditForm from './components/ModelEditForm';
-import Icon from '@/components/Iconfont';
 import { useQuery } from '@tanstack/react-query';
 import dynamic from 'next/dynamic';

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

 const ModelDetail = ({ modelId }: { modelId: string }) => {
  const { toast } = useToast();
  const router = useRouter();
  const { isPc, media } = useScreen();
  const { setLoading } = useGlobalStore();
-  const { openConfirm, ConfirmChild } = useConfirm({
-    content: '确认删除该模型?'
-  });
-  const SelectFileDom = useRef<HTMLInputElement>(null);
+
+  // const SelectFileDom = useRef<HTMLInputElement>(null);
  const [model, setModel] = useState<ModelSchema>(defaultModel);
  const formHooks = useForm<ModelSchema>({
    defaultValues: model
@@ -39,7 +29,7 @@ const ModelDetail = ({ modelId }: { modelId: string }) => {

  const canTrain = useMemo(() => {
    const openai = modelList.find((item) => item.model === model?.service.modelName);
-    return openai && openai.trainName;
+    return !!(openai && openai.trainName);
  }, [model]);

  /* 加载模型数据 */
@@ -91,34 +81,34 @@ const ModelDetail = ({ modelId }: { modelId: string }) => {
  }, [setLoading, model, router]);

  /* 上传数据集,触发微调 */
-  const startTraining = useCallback(
-    async (e: React.ChangeEvent<HTMLInputElement>) => {
-      if (!modelId || !e.target.files || e.target.files?.length === 0) return;
-      setLoading(true);
-      try {
-        const file = e.target.files[0];
-        const formData = new FormData();
-        formData.append('file', file);
-        await postTrainModel(modelId, formData);
+  // const startTraining = useCallback(
+  //   async (e: React.ChangeEvent<HTMLInputElement>) => {
+  //     if (!modelId || !e.target.files || e.target.files?.length === 0) return;
+  //     setLoading(true);
+  //     try {
+  //       const file = e.target.files[0];
+  //       const formData = new FormData();
+  //       formData.append('file', file);
+  //       await postTrainModel(modelId, formData);

-        toast({
-          title: '开始训练...',
-          status: 'success'
-        });
+  //       toast({
+  //         title: '开始训练...',
+  //         status: 'success'
+  //       });

-        // 重新获取模型
-        loadModel();
-      } catch (err: any) {
-        toast({
-          title: err?.message || '上传文件失败',
-          status: 'error'
-        });
-        console.log('error->', err);
-      }
-      setLoading(false);
-    },
-    [setLoading, loadModel, modelId, toast]
-  );
+  //       // 重新获取模型
+  //       loadModel();
+  //     } catch (err: any) {
+  //       toast({
+  //         title: err?.message || '上传文件失败',
+  //         status: 'error'
+  //       });
+  //       console.log('error->', err);
+  //     }
+  //     setLoading(false);
+  //   },
+  //   [setLoading, loadModel, modelId, toast]
+  // );

  /* 点击更新模型状态 */
  const handleClickUpdateStatus = useCallback(async () => {
@@ -250,87 +240,34 @@ const ModelDetail = ({ modelId }: { modelId: string }) => {
        )}
      </Card>
      <Grid mt={5} gridTemplateColumns={media('1fr 1fr', '1fr')} gridGap={5}>
-        <ModelEditForm formHooks={formHooks} />
+        <ModelEditForm formHooks={formHooks} handleDelModel={handleDelModel} canTrain={canTrain} />

-        {canTrain && (
+        {/* {canTrain && (
          <Card p={4}>
            <Training model={model} />
          </Card>
+        )} */}
+        {canTrain && model._id && (
+          <Card
+            p={4}
+            height={'700px'}
+            {...media(
+              {
+                gridColumnStart: 1,
+                gridColumnEnd: 3
+              },
+              {}
+            )}
+          >
+            <ModelDataCard model={model} />
+          </Card>
        )}
-
-        <Card p={4}>
-          <Box fontWeight={'bold'} fontSize={'lg'}>
-            神奇操作
-          </Box>
-          <Flex mt={5} alignItems={'center'}>
-            <Box flex={'0 0 80px'}>模型微调:</Box>
-            <Button
-              size={'sm'}
-              onClick={() => {
-                SelectFileDom.current?.click();
-              }}
-              title={!canTrain ? '模型不支持微调' : ''}
-              isDisabled={!canTrain}
-            >
-              上传数据集
-            </Button>
-            <Flex
-              as={'a'}
-              href="/TrainingTemplate.jsonl"
-              download
-              ml={5}
-              cursor={'pointer'}
-              alignItems={'center'}
-              color={'blue.500'}
-            >
-              <Icon name={'icon-yunxiazai'} color={'#3182ce'} />
-              下载模板
-            </Flex>
-          </Flex>
-          {/* 提示 */}
-          <Box mt={3} py={3} color={'blackAlpha.600'}>
-            <Box as={'li'} lineHeight={1.9}>
-              暂时需要使用自己的openai key
-            </Box>
-            <Box as={'li'} lineHeight={1.9}>
-              可以使用
-              <Box
-                as={'span'}
-                fontWeight={'bold'}
-                textDecoration={'underline'}
-                color={'blackAlpha.800'}
-                mx={2}
-                cursor={'pointer'}
-                onClick={() => router.push('/data/list')}
-              >
-                数据拆分
-              </Box>
-              功能，从任意文本中提取数据集。
-            </Box>
-            <Box as={'li'} lineHeight={1.9}>
-              每行包括一个 prompt 和一个 completion
-            </Box>
-            <Box as={'li'} lineHeight={1.9}>
-              prompt 必须以 {'</s>'} 结尾
-            </Box>
-            <Box as={'li'} lineHeight={1.9}>
-              completion 开头必须有一个空格，必须以 {'</s>'} 结尾
-            </Box>
-          </Box>
-          <Flex mt={5} alignItems={'center'}>
-            <Box flex={'0 0 80px'}>删除模型:</Box>
-            <Button colorScheme={'red'} size={'sm'} onClick={openConfirm(handleDelModel)}>
-              删除模型
-            </Button>
-          </Flex>
-        </Card>
      </Grid>

      {/* 文件选择 */}
-      <Box position={'absolute'} w={0} h={0} overflow={'hidden'}>
+      {/* <Box position={'absolute'} w={0} h={0} overflow={'hidden'}>
        <input ref={SelectFileDom} type="file" accept=".jsonl" onChange={startTraining} />
-      </Box>
-      <ConfirmChild />
+      </Box> */}
    </>
  );
 };
--- a/src/pages/model/list/components/CreateModel.tsx
+++ b/src/pages/model/list/components/CreateModel.tsx
--- a/src/pages/model/list/components/ModelPhoneList.tsx
+++ b/src/pages/model/list/components/ModelPhoneList.tsx
--- a/src/pages/model/list/components/ModelTable.tsx
+++ b/src/pages/model/list/components/ModelTable.tsx
--- a/src/pages/model/list/index.tsx
+++ b/src/pages/model/list/index.tsx