ChatGPT/FastGPT
1
0
Fork 0
mirror of https://github.com/labring/FastGPT.git synced 2025-07-23 05:12:39 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat: think tag parse (#3805) (#3808)

Browse Source 
* feat: think tag parse

* remove some model config

* feat: parse think tag test
This commit is contained in:
Archer
2025-02-17 20:57:36 +08:00
committed by GitHub
parent 4447e40364
commit 4d20274a97
10 changed files with 418 additions and 535 deletions
Download Patch File Download Diff File Expand all files Collapse all files

47
packages/service/core/ai/config.ts
View File

@@ -1,7 +1,9 @@
import OpenAI from '@fastgpt/global/core/ai';
import {
ChatCompletionCreateParamsNonStreaming,
ChatCompletionCreateParamsStreaming
ChatCompletionCreateParamsStreaming,
StreamChatType,
UnStreamChatType
} from '@fastgpt/global/core/ai/type';
import { getErrText } from '@fastgpt/global/common/error/utils';
import { addLog } from '../../common/system/log';
@@ -38,29 +40,30 @@ export const getAxiosConfig = (props?: { userKey?: OpenaiAccountType }) => {
};
};
type CompletionsBodyType =
| ChatCompletionCreateParamsNonStreaming
| ChatCompletionCreateParamsStreaming;
type InferResponseType<T extends CompletionsBodyType> =
T extends ChatCompletionCreateParamsStreaming
? OpenAI.Chat.Completions.ChatCompletionChunk
: OpenAI.Chat.Completions.ChatCompletion;
export const createChatCompletion = async <T extends CompletionsBodyType>({
export const createChatCompletion = async ({
body,
userKey,
timeout,
options
}: {
body: T;
body: ChatCompletionCreateParamsNonStreaming | ChatCompletionCreateParamsStreaming;
userKey?: OpenaiAccountType;
timeout?: number;
options?: OpenAI.RequestOptions;
}): Promise<{
response: InferResponseType<T>;
isStreamResponse: boolean;
getEmptyResponseTip: () => string;
}> => {
}): Promise<
{
getEmptyResponseTip: () => string;
} & (
| {
response: StreamChatType;
isStreamResponse: true;
}
| {
response: UnStreamChatType;
isStreamResponse: false;
}
)
> => {
try {
const modelConstantsData = getLLMModel(body.model);
@@ -96,9 +99,17 @@ export const createChatCompletion = async <T extends CompletionsBodyType>({
return i18nT('chat:LLM_model_response_empty');
};
if (isStreamResponse) {
return {
response,
isStreamResponse: true,
getEmptyResponseTip
};
}
return {
response: response as InferResponseType<T>,
isStreamResponse,
response,
isStreamResponse: false,
getEmptyResponseTip
};
} catch (error) {

461
packages/service/core/ai/config/provider/PPIO.json
View File

@@ -1,461 +1,4 @@
{
"provider": "PPIO",
"list": [
{
"model": "deepseek/deepseek-r1/community",
"name": "deepseek/deepseek-r1/community",
"maxContext": 64000,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "deepseek/deepseek-v3/community",
"name": "deepseek/deepseek-v3/community",
"maxContext": 64000,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "deepseek/deepseek-r1",
"name": "deepseek/deepseek-r1",
"maxContext": 64000,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "deepseek/deepseek-v3",
"name": "deepseek/deepseek-v3",
"maxContext": 64000,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "deepseek/deepseek-r1-distill-llama-70b",
"name": "deepseek/deepseek-r1-distill-llama-70b",
"maxContext": 32000,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "deepseek/deepseek-r1-distill-qwen-32b",
"name": "deepseek/deepseek-r1-distill-qwen-32b",
"maxContext": 64000,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "deepseek/deepseek-r1-distill-qwen-14b",
"name": "deepseek/deepseek-r1-distill-qwen-14b",
"maxContext": 64000,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "deepseek/deepseek-r1-distill-llama-8b",
"name": "deepseek/deepseek-r1-distill-llama-8b",
"maxContext": 32000,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "qwen/qwen-2.5-72b-instruct",
"name": "qwen/qwen-2.5-72b-instruct",
"maxContext": 32768,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "qwen/qwen-2-vl-72b-instruct",
"name": "qwen/qwen-2-vl-72b-instruct",
"maxContext": 32768,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "meta-llama/llama-3.2-3b-instruct",
"name": "meta-llama/llama-3.2-3b-instruct",
"maxContext": 32768,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "qwen/qwen2.5-32b-instruct",
"name": "qwen/qwen2.5-32b-instruct",
"maxContext": 32000,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "baichuan/baichuan2-13b-chat",
"name": "baichuan/baichuan2-13b-chat",
"maxContext": 14336,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "meta-llama/llama-3.1-70b-instruct",
"name": "meta-llama/llama-3.1-70b-instruct",
"maxContext": 32768,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "meta-llama/llama-3.1-8b-instruct",
"name": "meta-llama/llama-3.1-8b-instruct",
"maxContext": 32768,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "01-ai/yi-1.5-34b-chat",
"name": "01-ai/yi-1.5-34b-chat",
"maxContext": 16384,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "01-ai/yi-1.5-9b-chat",
"name": "01-ai/yi-1.5-9b-chat",
"maxContext": 16384,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "thudm/glm-4-9b-chat",
"name": "thudm/glm-4-9b-chat",
"maxContext": 32768,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
},
{
"model": "qwen/qwen-2-7b-instruct",
"name": "qwen/qwen-2-7b-instruct",
"maxContext": 32768,
"maxResponse": 8000,
"quoteMaxToken": 50000,
"maxTemperature": 2,
"vision": false,
"toolChoice": false,
"functionCall": false,
"defaultSystemChatPrompt": "",
"datasetProcess": true,
"usedInClassify": true,
"customCQPrompt": "",
"usedInExtractFields": true,
"usedInQueryExtension": true,
"customExtractPrompt": "",
"usedInToolCall": true,
"defaultConfig": {},
"fieldMap": {},
"type": "llm",
"showTopP": true,
"showStopSign": true
}
]
}
"list": []
}

25
packages/service/core/ai/utils.ts
View File

@@ -37,25 +37,26 @@ export const computedTemperature = ({
return temperature;
};
type CompletionsBodyType = (
type CompletionsBodyType =
| ChatCompletionCreateParamsNonStreaming
| ChatCompletionCreateParamsStreaming
) & {
response_format?: any;
json_schema?: string;
stop?: string;
};
| ChatCompletionCreateParamsStreaming;
type InferCompletionsBody<T> = T extends { stream: true }
? ChatCompletionCreateParamsStreaming
: ChatCompletionCreateParamsNonStreaming;
: T extends { stream: false }
? ChatCompletionCreateParamsNonStreaming
: ChatCompletionCreateParamsNonStreaming | ChatCompletionCreateParamsStreaming;
export const llmCompletionsBodyFormat = <T extends CompletionsBodyType>(
body: T,
body: T & {
response_format?: any;
json_schema?: string;
stop?: string;
},
model: string | LLMModelItemType
): InferCompletionsBody<T> => {
const modelData = typeof model === 'string' ? getLLMModel(model) : model;
if (!modelData) {
return body as InferCompletionsBody<T>;
return body as unknown as InferCompletionsBody<T>;
}
const response_format = body.response_format;
@@ -91,9 +92,7 @@ export const llmCompletionsBodyFormat = <T extends CompletionsBodyType>(
});
}
// console.log(requestBody);
return requestBody as InferCompletionsBody<T>;
return requestBody as unknown as InferCompletionsBody<T>;
};
export const llmStreamResponseToText = async (response: StreamChatType) => {

88
packages/service/core/workflow/dispatch/chat/oneapi.ts
View File

@@ -3,13 +3,13 @@ import { filterGPTMessageByMaxContext, loadRequestMessages } from '../../../chat
import type { ChatItemType, UserChatItemValueItemType } from '@fastgpt/global/core/chat/type.d';
import { ChatRoleEnum } from '@fastgpt/global/core/chat/constants';
import { SseResponseEventEnum } from '@fastgpt/global/core/workflow/runtime/constants';
import { textAdaptGptResponse } from '@fastgpt/global/core/workflow/runtime/utils';
import {
parseReasoningContent,
parseReasoningStreamContent,
textAdaptGptResponse
} from '@fastgpt/global/core/workflow/runtime/utils';
import { createChatCompletion } from '../../../ai/config';
import type {
ChatCompletion,
ChatCompletionMessageParam,
StreamChatType
} from '@fastgpt/global/core/ai/type.d';
import type { ChatCompletionMessageParam, StreamChatType } from '@fastgpt/global/core/ai/type.d';
import { formatModelChars2Points } from '../../../../support/wallet/usage/utils';
import type { LLMModelItemType } from '@fastgpt/global/core/ai/model.d';
import { postTextCensor } from '../../../../common/api/requestPlusApi';
@@ -195,7 +195,13 @@ export const dispatchChatCompletion = async (props: ChatProps): Promise<ChatResp
});
const { answerText, reasoningText } = await (async () => {
if (res && isStreamResponse) {
if (isStreamResponse) {
if (!res) {
return {
answerText: '',
reasoningText: ''
};
}
// sse response
const { answer, reasoning } = await streamResponse({
res,
@@ -210,34 +216,49 @@ export const dispatchChatCompletion = async (props: ChatProps): Promise<ChatResp
reasoningText: reasoning
};
} else {
const unStreamResponse = response as ChatCompletion;
const answer = unStreamResponse.choices?.[0]?.message?.content || '';
// @ts-ignore
const reasoning = unStreamResponse.choices?.[0]?.message?.reasoning_content || '';
const { content, reasoningContent } = (() => {
const content = response.choices?.[0]?.message?.content || '';
// @ts-ignore
const reasoningContent: string = response.choices?.[0]?.message?.reasoning_content || '';
// API already parse reasoning content
if (reasoningContent || !aiChatReasoning) {
return {
content,
reasoningContent
};
}
const [think, answer] = parseReasoningContent(content);
return {
content: answer,
reasoningContent: think
};
})();
// Some models do not support streaming
if (stream) {
if (isResponseAnswerText && answer) {
if (aiChatReasoning && reasoningContent) {
workflowStreamResponse?.({
event: SseResponseEventEnum.fastAnswer,
data: textAdaptGptResponse({
text: answer
reasoning_content: reasoningContent
})
});
}
if (aiChatReasoning && reasoning) {
if (isResponseAnswerText && content) {
workflowStreamResponse?.({
event: SseResponseEventEnum.fastAnswer,
data: textAdaptGptResponse({
reasoning_content: reasoning
text: content
})
});
}
}
return {
answerText: answer,
reasoningText: reasoning
answerText: content,
reasoningText: reasoningContent
};
}
})();
@@ -267,7 +288,7 @@ export const dispatchChatCompletion = async (props: ChatProps): Promise<ChatResp
});
return {
answerText,
answerText: answerText.trim(),
reasoningText,
[DispatchNodeResponseKeyEnum.nodeResponse]: {
totalPoints: externalProvider.openaiAccount?.key ? 0 : totalPoints,
@@ -500,26 +521,18 @@ async function streamResponse({
});
let answer = '';
let reasoning = '';
const { parsePart, getStartTagBuffer } = parseReasoningStreamContent();
for await (const part of stream) {
if (res.closed) {
stream.controller?.abort();
break;
}
const content = part.choices?.[0]?.delta?.content || '';
const [reasoningContent, content] = parsePart(part, aiChatReasoning);
answer += content;
if (isResponseAnswerText && content) {
workflowStreamResponse?.({
write,
event: SseResponseEventEnum.answer,
data: textAdaptGptResponse({
text: content
})
});
}
const reasoningContent = part.choices?.[0]?.delta?.reasoning_content || '';
reasoning += reasoningContent;
if (aiChatReasoning && reasoningContent) {
workflowStreamResponse?.({
write,
@@ -529,6 +542,21 @@ async function streamResponse({
})
});
}
if (isResponseAnswerText && content) {
workflowStreamResponse?.({
write,
event: SseResponseEventEnum.answer,
data: textAdaptGptResponse({
text: content
})
});
}
}
// if answer is empty, try to get value from startTagBuffer. (Cause: The response content is too short to exceed the minimum parse length)
if (answer === '') {
answer = getStartTagBuffer();
}
return { answer, reasoning };

17
packages/service/core/workflow/dispatch/index.ts
View File

@@ -243,12 +243,17 @@ export async function dispatchWorkFlow(data: Props): Promise<DispatchFlowRespons
chatAssistantResponse = chatAssistantResponse.concat(assistantResponses);
} else {
if (reasoningText) {
chatAssistantResponse.push({
type: ChatItemValueTypeEnum.reasoning,
reasoning: {
content: reasoningText
}
});
const isResponseReasoningText = inputs.find(
(item) => item.key === NodeInputKeyEnum.aiChatReasoning
)?.value;
if (isResponseReasoningText) {
chatAssistantResponse.push({
type: ChatItemValueTypeEnum.reasoning,
reasoning: {
content: reasoningText
}
});
}
}
if (answerText) {
// save assistant text response