mirror of
https://github.com/labring/FastGPT.git
synced 2025-07-22 20:37:48 +00:00
4.7-production (#1053)
* 4.7-alpha3 (#62) * doc * Optimize possible null Pointers and parts of Ux * fix: mulity index training error * feat: doc and rename question guide * fix ios speech input (#59) * fix: prompt editor variables nowrap (#61) * change openapi import in http module with curl import (#60) * chore(ui): dataset import modal ui (#58) * chore(ui): dataset import modal ui * use component * fix height * 4.7 (#63) * fix: claude3 image type verification failed (#1038) (#1040) * perf: curl import modal * doc img * perf: adapt cohere rerank * perf: code * perf: input style * doc --------- Co-authored-by: xiaotian <dimsky@163.com> * fix: ts * docker deploy * perf: prompt call * doc * ts * finish ui * perf: outlink detail ux * perf: user schema * fix: plugin update * feat: get current time plugin * fix: ts * perf: fetch anamation * perf: mark ux * doc * perf: select app ux * fix: split text custom string conflict * peref: inform readed * doc * memo flow component * perf: version * faq * feat: flow max runtimes * feat: similarity tip * feat: auto detect file encoding * Supports asymmetric vector model * fix: ts * perf: max w * move code * perf: hide whisper * fix: ts * feat: system msg modal * perf: catch error * perf: inform tip * fix: inform --------- Co-authored-by: heheer <71265218+newfish-cmyk@users.noreply.github.com> Co-authored-by: xiaotian <dimsky@163.com>
This commit is contained in:
Archer
@@ -1,14 +1,16 @@
|
||||
import { VectorModelItemType } from '@fastgpt/global/core/ai/model.d';
|
||||
import { getAIApi } from '../config';
|
||||
import { countPromptTokens } from '@fastgpt/global/common/string/tiktoken';
|
||||
import { EmbeddingTypeEnm } from '@fastgpt/global/core/ai/constants';
|
||||
|
||||
type GetVectorProps = {
|
||||
model: VectorModelItemType;
|
||||
input: string;
|
||||
type?: `${EmbeddingTypeEnm}`;
|
||||
};
|
||||
|
||||
// text to vector
|
||||
export async function getVectorsByText({ model, input }: GetVectorProps) {
|
||||
export async function getVectorsByText({ model, input, type }: GetVectorProps) {
|
||||
if (!input) {
|
||||
return Promise.reject({
|
||||
code: 500,
|
||||
@@ -23,6 +25,8 @@ export async function getVectorsByText({ model, input }: GetVectorProps) {
|
||||
const result = await ai.embeddings
|
||||
.create({
|
||||
...model.defaultConfig,
|
||||
...(type === EmbeddingTypeEnm.db && model.dbConfig),
|
||||
...(type === EmbeddingTypeEnm.query && model.queryConfig),
|
||||
model: model.model,
|
||||
input: [input]
|
||||
})
|
||||
|
||||
@@ -160,7 +160,7 @@ A: ${chatBg}
|
||||
|
||||
return {
|
||||
rawQuery: query,
|
||||
extensionQueries: queries,
|
||||
extensionQueries: Array.isArray(queries) ? queries : [],
|
||||
model,
|
||||
tokens: countGptMessagesTokens(messages)
|
||||
};
|
||||
|
||||
@@ -1,7 +1,21 @@
|
||||
import { PostReRankProps, PostReRankResponse } from '@fastgpt/global/core/ai/api.d';
|
||||
import { POST } from '../../../common/api/serverRequest';
|
||||
|
||||
export function reRankRecall({ query, inputs }: PostReRankProps) {
|
||||
type PostReRankResponse = {
|
||||
id: string;
|
||||
results: {
|
||||
index: number;
|
||||
relevance_score: number;
|
||||
}[];
|
||||
};
|
||||
type ReRankCallResult = { id: string; score?: number }[];
|
||||
|
||||
export function reRankRecall({
|
||||
query,
|
||||
documents
|
||||
}: {
|
||||
query: string;
|
||||
documents: { id: string; text: string }[];
|
||||
}): Promise<ReRankCallResult> {
|
||||
const model = global.reRankModels[0];
|
||||
|
||||
if (!model || !model?.requestUrl) {
|
||||
@@ -12,19 +26,24 @@ export function reRankRecall({ query, inputs }: PostReRankProps) {
|
||||
return POST<PostReRankResponse>(
|
||||
model.requestUrl,
|
||||
{
|
||||
model: model.model,
|
||||
query,
|
||||
inputs
|
||||
documents: documents.map((doc) => doc.text)
|
||||
},
|
||||
{
|
||||
headers: {
|
||||
Authorization: `Bearer ${model.requestAuth}`
|
||||
},
|
||||
timeout: 120000
|
||||
timeout: 30000
|
||||
}
|
||||
)
|
||||
.then((data) => {
|
||||
console.log('rerank time:', Date.now() - start);
|
||||
return data;
|
||||
|
||||
return data?.results?.map((item) => ({
|
||||
id: documents[item.index].id,
|
||||
score: item.relevance_score
|
||||
}));
|
||||
})
|
||||
.catch((err) => {
|
||||
console.log('rerank error:', err);
|
||||
|
||||
@@ -77,7 +77,8 @@ export async function searchDatasetData(props: SearchDatasetDataProps) {
|
||||
const embeddingRecall = async ({ query, limit }: { query: string; limit: number }) => {
|
||||
const { vectors, tokens } = await getVectorsByText({
|
||||
model: getVectorModel(model),
|
||||
input: query
|
||||
input: query,
|
||||
type: 'query'
|
||||
});
|
||||
|
||||
const { results } = await recallFromVectorStore({
|
||||
@@ -225,7 +226,7 @@ export async function searchDatasetData(props: SearchDatasetDataProps) {
|
||||
try {
|
||||
const results = await reRankRecall({
|
||||
query,
|
||||
inputs: data.map((item) => ({
|
||||
documents: data.map((item) => ({
|
||||
id: item.id,
|
||||
text: `${item.q}\n${item.a}`
|
||||
}))
|
||||
|
||||
@@ -1,12 +1,9 @@
|
||||
export const Prompt_Tool_Call = `<Instruction>
|
||||
你是一个智能机器人，除了可以回答用户问题外，你还掌握工具的使用能力。有时候，你可以依赖工具的运行结果，来更准确的回答用户。
|
||||
下面是你可以使用的工具，使用 JSON Schema 的格式声明，其中 toolId 是工具的 description 是工具的描述，parameters 是工具的参数，包括参数的类型和描述，required 是必填参数的列表。
|
||||
|
||||
"""
|
||||
{{toolsPrompt}}
|
||||
"""
|
||||
工具使用了 JSON Schema 的格式声明，其中 toolId 是工具的 description 是工具的描述，parameters 是工具的参数，包括参数的类型和描述，required 是必填参数的列表。
|
||||
|
||||
接下来，请你根据工具描述，决定回答问题或是使用工具。在完成任务过程中，USER代表用户的输入，TOOL_RESPONSE代表工具运行结果。ASSISTANT 代表你的输出。
|
||||
请你根据工具描述，决定回答问题或是使用工具。在完成任务过程中，USER代表用户的输入，TOOL_RESPONSE代表工具运行结果。ASSISTANT 代表你的输出。
|
||||
你的每次输出都必须以0,1开头，代表是否需要调用工具：
|
||||
0: 不使用工具，直接回答内容。
|
||||
1: 使用工具，返回工具调用的参数。
|
||||
@@ -29,7 +26,13 @@ TOOL_RESPONSE: """
|
||||
ANSWER: 0: 今天杭州是晴天，适合去西湖、灵隐寺、千岛湖等地玩。
|
||||
</Instruction>
|
||||
|
||||
现在，我们开始吧！
|
||||
现在，我们开始吧！下面是你本次可以使用的工具：
|
||||
|
||||
"""
|
||||
{{toolsPrompt}}
|
||||
"""
|
||||
|
||||
下面是正式的对话内容：
|
||||
|
||||
USER: {{question}}
|
||||
ANSWER:
|
||||
|
||||
@@ -125,6 +125,7 @@ export async function dispatchWorkFlow({
|
||||
}
|
||||
if (nodeDispatchUsages) {
|
||||
chatNodeUsages = chatNodeUsages.concat(nodeDispatchUsages);
|
||||
props.maxRunTimes -= nodeDispatchUsages.length;
|
||||
}
|
||||
if (toolResponses !== undefined) {
|
||||
if (Array.isArray(toolResponses) && toolResponses.length === 0) return;
|
||||
@@ -217,7 +218,7 @@ export async function dispatchWorkFlow({
|
||||
);
|
||||
}
|
||||
async function moduleRun(module: RunningModuleItemType): Promise<any> {
|
||||
if (res.closed) return Promise.resolve();
|
||||
if (res.closed || props.maxRunTimes <= 0) return Promise.resolve();
|
||||
|
||||
if (stream && detail && module.showStatus) {
|
||||
responseStatus({
|
||||
|
||||
@@ -19,7 +19,6 @@ export const dispatchRunPlugin = async (props: RunPluginProps): Promise<RunPlugi
|
||||
mode,
|
||||
teamId,
|
||||
tmbId,
|
||||
module,
|
||||
params: { pluginId, ...data }
|
||||
} = props;
|
||||
|
||||
|
||||
@@ -157,7 +157,7 @@ async function fetchData({
|
||||
body: Record<string, any>;
|
||||
params: Record<string, any>;
|
||||
}): Promise<Record<string, any>> {
|
||||
const { data: response } = await axios<Record<string, any>>({
|
||||
const { data: response } = await axios({
|
||||
method,
|
||||
baseURL: `http://${SERVICE_LOCAL_HOST}`,
|
||||
url,
|
||||
@@ -241,7 +241,8 @@ async function fetchData({
|
||||
};
|
||||
|
||||
return {
|
||||
formatResponse: parseJson(response),
|
||||
formatResponse:
|
||||
typeof response === 'object' && !Array.isArray(response) ? parseJson(response) : {},
|
||||
rawResponse: response
|
||||
};
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user