feat(data-process): 支持思维链输出类型

This commit is contained in:
caoxiaozhu
2026-07-27 13:08:44 +08:00
parent ecafb7eb13
commit de2e8952b5
14 changed files with 443 additions and 46 deletions

View File

@@ -9,7 +9,7 @@ import type {
} from './types'
import { normalizeQaPairsGenerationCount } from './types'
export const DEFAULT_GENERATION_PROMPT = '你是一名专业的数据生成助手。请根据输入内容生成准确、完整、可直接用于模型训练的问答数据。仅输出符合目标格式的内容,答案应事实清晰、语言自然,不要添加分析过程、说明或无关内容。'
export const DEFAULT_GENERATION_PROMPT = '你是一名专业的数据生成助手。请根据输入内容生成准确、完整、可直接用于模型训练的问答数据。仅输出符合所选输出类型和目标格式的内容,答案应事实清晰、语言自然,不要添加无关说明。'
export function createDefaultStructuredOptions(): StructuredProcessOptions {
return {
@@ -19,6 +19,7 @@ export function createDefaultStructuredOptions(): StructuredProcessOptions {
datasetSplit: { train: 80, validation: 10, test: 10 },
generationModelId: '',
generationPrompt: DEFAULT_GENERATION_PROMPT,
outputType: 'standard',
temperature: 0.7,
maxTokens: 1024,
jsonMode: false,
@@ -52,6 +53,7 @@ export function createDefaultUnstructuredOptions(): UnstructuredProcessOptions {
datasetSplit: { train: 80, validation: 10, test: 10 },
generationModelId: '',
generationPrompt: DEFAULT_GENERATION_PROMPT,
outputType: 'standard',
temperature: 0.7,
maxTokens: 1024,
jsonMode: false,
@@ -93,6 +95,9 @@ function generationOptionsFromConfig(
return {
generationModelId: configValue(config, 'generation_model_id', defaults.generationModelId),
generationPrompt: String(configValue(config, 'generation_prompt', defaults.generationPrompt)),
outputType: configValue(config, 'output_type', defaults.outputType) === 'reasoning'
? 'reasoning'
: 'standard',
temperature: numberValue(config, 'temperature', defaults.temperature),
maxTokens: numberValue(config, 'max_tokens', defaults.maxTokens),
jsonMode: Boolean(configValue(config, 'json_mode', defaults.jsonMode)),
@@ -218,6 +223,7 @@ export function generationAffectingOptionsFor(
datasetSplit: options.datasetSplit,
generationModelId: options.generationModelId,
generationPrompt: options.generationPrompt,
outputType: options.outputType,
temperature: options.temperature,
maxTokens: options.maxTokens,
jsonMode: options.jsonMode,