jyapayne/dify
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

feat: add retriever rank fe (#1557)

Browse source 
Co-authored-by: StyleZhang <jasonapring2015@outlook.com>
This commit is contained in:
Joel 2023-11-18 11:53:35 +08:00 • committed by GitHub
commit 888e8c6dac
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
80 changed files with 2757 additions and 467 deletions
Download patch file Download diff file Expand all files Collapse all files

11
web/i18n/lang/app-debug.en.ts
View file

@ -305,11 +305,22 @@ const translation = {
},
result: 'Output Text',
datasetConfig: {
settingTitle: 'Retrieve Settings',
retrieveOneWay: {
title: 'N-to-1 retrieval',
description: 'Based on user intent and dataset descriptions, the Agent autonomously selects the best dataset for querying. Best for applications with distinct, limited datasets.',
},
retrieveMultiWay: {
title: 'Multi-path retrieval',
description: 'Based on user intent, queries across all datasets, retrieves relevant text from multi-sources, and selects the best results matching the user query after reranking. Configuration of the Rerank model API is required.',
},
rerankModelRequired: 'Rerank model is required',
params: 'Params',
top_k: 'Top K',
top_kTip: 'Used to filter segments that are most similar to user questions. The system will also dynamically adjust the value of Top K, according to max_tokens of the selected model.',
score_threshold: 'Score Threshold',
score_thresholdTip: 'Used to set the similarity threshold for segment filtering.',
retrieveChangeTip: 'Modifying the index mode and retrieval mode may affect applications associated with this dataset.',
},
}

11
web/i18n/lang/app-debug.zh.ts
View file

@ -299,11 +299,22 @@ const translation = {
},
result: '结果',
datasetConfig: {
settingTitle: '召回设置',
retrieveOneWay: {
title: 'N选1召回',
description: '根据用户意图和数据集描述，由 Agent 自主判断选择最匹配的单个数据集来查询相关文本，适合数据集区分度大且数据集数量偏少的应用。',
},
retrieveMultiWay: {
title: '多路召回',
description: '根据用户意图同时匹配所有数据集，从多路数据集查询相关文本片段，经过重排序步骤，从多路查询结果中选择匹配用户问题的最佳结果，需配置 Rerank 模型 API。',
},
rerankModelRequired: '请选择 Rerank 模型',
params: '参数设置',
top_k: 'Top K',
top_kTip: '用于筛选与用户问题相似度最高的文本片段。系统同时会根据选用模型上下文窗口大小动态调整分段数量。',
score_threshold: 'Score 阈值',
score_thresholdTip: '用于设置文本片段筛选的相似度阈值。',
retrieveChangeTip: '修改索引模式和检索模式可能会影响与该数据集关联的应用程序。',
},
}

14
web/i18n/lang/common.en.ts
View file

@ -223,6 +223,8 @@ const translation = {
},
},
modelProvider: {
systemModelSettings: 'System Model Settings',
systemModelSettingsLink: 'Why is it necessary to set up a system model?',
selectModel: 'Select your model',
setupModelFirst: 'Please set up your model first',
systemReasoningModel: {
@ -237,6 +239,10 @@ const translation = {
key: 'Speech-to-Text Model',
tip: 'Set the default model for speech-to-text input in conversation.',
},
rerankModel: {
key: 'Rerank Model',
tip: 'Rerank model will reorder the candidate document list based on the semantic match with user query, improving the results of semantic ranking',
},
quota: 'Quota',
searchModel: 'Search model',
noModelFound: 'No model found for {{model}}',
@ -244,6 +250,8 @@ const translation = {
showMoreModelProvider: 'Show more model provider',
selector: {
tip: 'This model has been removed. Please add a model or select another model.',
emptyTip: 'No available models',
emptySetting: 'Please go to settings to configure',
},
card: {
quota: 'QUOTA',
@ -351,7 +359,7 @@ const translation = {
},
datasetMenus: {
documents: 'Documents',
hitTesting: 'Hit Testing',
hitTesting: 'Retrieval Testing',
settings: 'Settings',
emptyTip: 'The data set has not been associated, please go to the application or plug-in to complete the association.',
viewDoc: 'View documentation',
@ -382,9 +390,9 @@ const translation = {
title: 'CITATIONS',
linkToDataset: 'Link to dataset',
characters: 'Characters:',
hitCount: 'Hit count:',
hitCount: 'Retrieval count:',
vectorHash: 'Vector hash:',
hitScore: 'Hit Score:',
hitScore: 'Retrieval Score:',
},
},
promptEditor: {

14
web/i18n/lang/common.zh.ts
View file

@ -223,6 +223,8 @@ const translation = {
},
},
modelProvider: {
systemModelSettings: '系统模型设置',
systemModelSettingsLink: '为什么需要设置系统模型？',
selectModel: '选择您的模型',
setupModelFirst: '请先设置您的模型',
systemReasoningModel: {
@ -237,6 +239,10 @@ const translation = {
key: '语音转文本模型',
tip: '设置对话中语音转文字输入的默认使用模型。',
},
rerankModel: {
key: 'Rerank 模型',
tip: '重排序模型将根据候选文档列表与用户问题语义匹配度进行重新排序，从而改进语义排序的结果',
},
quota: '额度',
searchModel: '搜索模型',
noModelFound: '找不到模型 {{model}}',
@ -244,6 +250,8 @@ const translation = {
showMoreModelProvider: '显示更多模型提供商',
selector: {
tip: '该模型已被删除。请添模型或选择其他模型。',
emptyTip: '无可用模型',
emptySetting: '请前往设置进行配置',
},
card: {
quota: '额度',
@ -351,7 +359,7 @@ const translation = {
},
datasetMenus: {
documents: '文档',
hitTesting: '命中测试',
hitTesting: '召回测试',
settings: '设置',
emptyTip: ' 数据集尚未关联，请前往应用程序或插件完成关联。',
viewDoc: '查看文档',
@ -382,9 +390,9 @@ const translation = {
title: '引用',
linkToDataset: '跳转至数据集',
characters: '字符：',
hitCount: '命中次数：',
hitCount: '召回次数：',
vectorHash: '向量哈希：',
hitScore: '命中得分：',
hitScore: '召回得分：',
},
},
promptEditor: {

1
web/i18n/lang/dataset-creation.en.ts
View file

@ -101,6 +101,7 @@ const translation = {
previewSwitchTipEnd: ' consume additional tokens',
characters: 'characters',
indexSettedTip: 'To change the index method, please go to the ',
retrivalSettedTip: 'To change the index method, please go to the ',
datasetSettingLink: 'dataset settings.',
},
stepThree: {

1
web/i18n/lang/dataset-creation.zh.ts
View file

@ -101,6 +101,7 @@ const translation = {
previewSwitchTipEnd: '消耗额外的 token',
characters: '字符',
indexSettedTip: '要更改索引方法，请转到',
retrivalSettedTip: '要更改检索方法，请转到',
datasetSettingLink: '数据集设置。',
},
stepThree: {

6
web/i18n/lang/dataset-documents.en.ts
View file

@ -8,7 +8,7 @@ const translation = {
header: {
fileName: 'FILE NAME',
words: 'WORDS',
hitCount: 'HIT COUNT',
hitCount: 'RETRIEVAL COUNT',
uploadTime: 'UPLOAD TIME',
status: 'STATUS',
action: 'ACTION',
@ -216,7 +216,7 @@ const translation = {
segmentLength: 'Segment length',
avgParagraphLength: 'Avg. paragraph length',
paragraphs: 'Paragraphs',
hitCount: 'Hit count',
hitCount: 'Retrieval count',
embeddingTime: 'Embedding time',
embeddedSpend: 'Embedded spend',
},
@ -332,7 +332,7 @@ const translation = {
addKeyWord: 'Add key word',
keywordError: 'The maximum length of keyword is 20',
characters: 'characters',
hitCount: 'hit count',
hitCount: 'Retrieval count',
vectorHash: 'Vector hash: ',
questionPlaceholder: 'add question here',
questionEmpty: 'Question can not be empty',

6
web/i18n/lang/dataset-documents.zh.ts
View file

@ -8,7 +8,7 @@ const translation = {
header: {
fileName: '文件名',
words: '字符数',
hitCount: '命中次数',
hitCount: '召回次数',
uploadTime: '上传时间',
status: '状态',
action: '操作',
@ -215,7 +215,7 @@ const translation = {
segmentLength: '段落长度',
avgParagraphLength: '平均段落长度',
paragraphs: '段落数量',
hitCount: '命中次数',
hitCount: '召回次数',
embeddingTime: '嵌入时间',
embeddedSpend: '嵌入花费',
},
@ -331,7 +331,7 @@ const translation = {
addKeyWord: '添加关键词',
keywordError: '关键词最大长度为 20',
characters: '字符',
hitCount: '命中次数',
hitCount: '召回次数',
vectorHash: '向量哈希：',
questionPlaceholder: '在这里添加问题',
questionEmpty: '问题不能为空',

18
web/i18n/lang/dataset-hit-testing.en.ts
View file

@ -1,13 +1,13 @@
const translation = {
title: "Hit Testing",
desc: "Test the hitting effect of the dataset based on the given query text.",
title: 'Retrieval Testing',
desc: 'Test the hitting effect of the dataset based on the given query text.',
dateTimeFormat: 'MM/DD/YYYY hh:mm A',
recents: 'Recents',
table: {
header: {
source: "Source",
text: "Text",
time: "Time",
source: 'Source',
text: 'Text',
time: 'Time',
},
},
input: {
@ -18,11 +18,11 @@ const translation = {
testing: 'Testing',
},
hit: {
title: "HIT PARAGRAPHS",
emptyTip: 'Hit Testing results will show here',
title: 'RETRIEVAL PARAGRAPHS',
emptyTip: 'Retrieval Testing results will show here',
},
noRecentTip: 'No recent query results here',
viewChart: 'View VECTOR CHART',
};
}
export default translation;
export default translation

14
web/i18n/lang/dataset-hit-testing.zh.ts
View file

@ -1,13 +1,13 @@
const translation = {
title: '命中测试',
desc: '基于给定的查询文本测试数据集的命中效果。',
title: '召回测试',
desc: '基于给定的查询文本测试数据集的召回效果。',
dateTimeFormat: 'YYYY-MM-DD HH:mm',
recents: '最近查询',
table: {
header: {
source: "数据源",
text: "文本",
time: "时间",
source: '数据源',
text: '文本',
time: '时间',
},
},
input: {
@ -18,8 +18,8 @@ const translation = {
testing: '测试',
},
hit: {
title: "命中段落",
emptyTip: '命中测试结果将展示在这里',
title: '召回段落',
emptyTip: '召回测试结果将展示在这里',
},
noRecentTip: '最近无查询结果',
viewChart: '查看向量图表',

6
web/i18n/lang/dataset-settings.en.ts
View file

@ -20,6 +20,12 @@ const translation = {
embeddingModel: 'Embedding Model',
embeddingModelTip: 'Change the embedded model, please go to ',
embeddingModelTipLink: 'Settings',
retrievalSetting: {
title: 'Retrieval setting',
learnMore: 'Learn more',
description: ' about retrieval method.',
longDescription: ' about retrieval method, you can change this at any time in the dataset settings.',
},
save: 'Save',
},
}

6
web/i18n/lang/dataset-settings.zh.ts
View file

@ -20,6 +20,12 @@ const translation = {
embeddingModel: 'Embedding 模型',
embeddingModelTip: '修改 Embedding 模型，请去',
embeddingModelTipLink: '设置',
retrievalSetting: {
title: '检索设置',
learnMore: '了解更多',
description: '关于检索方法。',
longDescription: '关于检索方法，您可以随时在数据集设置中更改此设置。',
},
save: '保存',
},
}

21
web/i18n/lang/dataset.en.ts
View file

@ -20,6 +20,27 @@ const translation = {
unavailableTip: 'Embedding model is not available, the default embedding model needs to be configured',
datasets: 'DATASETS',
datasetsApi: 'API',
retrieval: {
semantic_search: {
title: 'Vector Search',
description: 'Generate query embeddings and search for the text chunk most similar to its vector representation.',
},
full_text_search: {
title: 'Full-Text Search',
description: 'Index all terms in the document, allowing users to search any term and retrieve relevant text chunk containing those terms.',
},
hybrid_search: {
title: 'Hybrid Search',
description: 'Execute full-text search and vector searches simultaneously, re-rank to select the best match for the user\'s query. Configuration of the Rerank model APIis necessary.',
recommend: 'Recommend',
},
invertedIndex: {
title: 'Inverted Index',
description: 'Inverted Index is a structure used for efficient retrieval. Organized by terms, each term points to documents or web pages containing it.',
},
change: 'Change',
changeRetrievalMethod: 'Change retrieval method',
},
}
export default translation

21
web/i18n/lang/dataset.zh.ts
View file

@ -20,6 +20,27 @@ const translation = {
unavailableTip: '由于 embedding 模型不可用，需要配置默认 embedding 模型',
datasets: '数据集',
datasetsApi: 'API',
retrieval: {
semantic_search: {
title: '向量检索',
description: '通过生成查询嵌入并查询与其向量表示最相似的文本分段',
},
full_text_search: {
title: '全文检索',
description: '索引文档中的所有词汇，从而允许用户查询任意词汇，并返回包含这些词汇的文本片段',
},
hybrid_search: {
title: '混合检索',
description: '同时执行全文检索和向量检索，并应用重排序步骤，从两类查询结果中选择匹配用户问题的最佳结果，需配置 Rerank 模型 API',
recommend: '推荐',
},
invertedIndex: {
title: '倒排索引',
description: '倒排索引是一种用于高效检索的结构。按术语组织，每个术语指向包含它的文档或网页',
},
change: '更改',
changeRetrievalMethod: '更改检索方法',
},
}
export default translation