'Final step! After success, leave the rest to Infiniflow AI.',
topK:'Top K',
topKTip:
"For the computaion cost, not all the retrieved chunk will be computed vector cosine similarity with query. The bigger the 'Top K' is, the higher the recall rate is, the slower the retrieval speed is.",
similarityThreshold:'Similarity threshold',
similarityThresholdTip:
"We use hybrid similarity score to evaluate distance between two lines of text. It's weighted keywords similarity and vector cosine similarity. If the similarity between query and chunk is less than this threshold, the chunk will be filtered out.",
"We use hybrid similarity score to evaluate distance between two lines of text. It's weighted keywords similarity and vector cosine similarity. The sum of both weights is 1.0.",
testText:'Test text',
testTextPlaceholder:'Please input your question!',
'page ranges: Define the page ranges that need to be parsed. The pages that not included in these ranges will be ignored.',
fromPlaceholder:'from',
fromMessage:'Missing start page number',
toPlaceholder:'to',
toMessage:'Missing end page number(excluded)',
layoutRecognize:'Layout recognize',
layoutRecognizeTip:
'Use visual models for layout analysis to better identify document structure, find where the titles, text blocks, images, and tables are. Without this feature, only the plain text of the PDF can be obtained.',
taskPageSize:'Task page size',
taskPageSizeMessage:'Please input your task page size!',
taskPageSizeTip:`If using layout recognize, the PDF file will be split into groups of successive. Layout analysis will be performed parallelly between groups to increase the processing speed. The 'Task page size' determines the size of groups. The larger the page size is, the lower the chance of splitting continuous text between pages into different chunks.`,
addPage:'Add page',
greaterThan:'The current value must be greater than to!',
greaterThanPrevious:
'The current value must be greater than the previous to!',
selectFiles:'Select files',
changeSpecificCategory:'Change specific category',
uploadTitle:'Click or drag file to this area to upload',
uploadDescription:
'Support for a single or bulk upload. Strictly prohibited from uploading company data or other banned files.',
'Update your knowledge base details especially parsing method here.',
name:'Knowledge base name',
photo:'Knowledge base photo',
description:'Description',
language:'Language',
languageMessage:'Please input your language!',
languagePlaceholder:'Please input your language!',
permissions:'Permissions',
embeddingModel:'Embedding model',
chunkTokenNumber:'Chunk token number',
chunkTokenNumberMessage:'Chunk token number is required',
embeddingModelTip:
"The embedding model used to embedding chunks. It's unchangable once the knowledgebase has chunks. You need to delete all the chunks if you want to change it.",
permissionsTip:
"If the permission is 'Team', all the team member can manipulate the knowledgebase.",
chunkTokenNumberTip:
'It determine the token number of a chunk approximately.',
chunkMethod:'Chunk method',
chunkMethodTip:'The instruction is at right.',
upload:'Upload',
english:'English',
chinese:'Chinese',
embeddingModelPlaceholder:'Please select a embedding model',
chunkMethodPlaceholder:'Please select a chunk method',
emptyResponseTip:`If nothing is retrieved with user's question in the knowledgebase, it will use this as an answer. If you want LLM comes up with its own opinion when nothing is retrieved, leave this blank.`,
setAnOpener:'Set an opener',
setAnOpenerInitial:`Hi! I'm your assistant, what can I do for you?`,
setAnOpenerTip:'How do you want to welcome your clients?',
systemInitialValue:`You are an intelligent assistant. Please summarize the content of the knowledge base to answer the question. Please list the data in the knowledge base and answer in detail. When all knowledge base content is irrelevant to the question, your answer must include the sentence "The answer you are looking for is not found in the knowledge base!" Answers need to consider chat history.
Hereistheknowledgebase:
{knowledge}
Theaboveistheknowledgebase.`,
systemMessage:'Please input!',
systemTip:
'Instructions you need LLM to follow when LLM answers questions, like charactor design, answer length and answer language etc.',
topN:'Top N',
topNTip:`Not all the chunks whose similarity score is above the 'simialrity threashold' will be feed to LLMs. LLM can only see these 'Top N' chunks.`,
variable:'Variable',
variableTip:`If you use dialog APIs, the varialbes might help you chat with your clients with different strategies.
freedomTip:`'Precise' means the LLM will be conservative and answer your question cautiously. 'Improvise' means the you want LLM talk much and freely. 'Balance' is between cautiously and freely.`,
temperature:'Temperature',
temperatureMessage:'Temperature is required',
temperatureTip:
'This parameter controls the randomness of predictions by the model. A lower temperature makes the model more confident in its responses, while a higher temperature makes it more creative and diverse.',
topP:'Top P',
topPMessage:'Top P is required',
topPTip:
'Also known as “nucleus sampling,” this parameter sets a threshold to select a smaller set of words to sample from. It focuses on the most likely words, cutting off the less probable ones.',
presencePenalty:'Presence Penalty',
presencePenaltyMessage:'Presence Penalty is required',
presencePenaltyTip:
'This discourages the model from repeating the same information by penalizing words that have already appeared in the conversation.',
frequencyPenalty:'Frequency Penalty',
frequencyPenaltyMessage:'Frequency Penalty is required',
frequencyPenaltyTip:
'Similar to the presence penalty, this reduces the model’s tendency to repeat the same words frequently.',
maxTokens:'Max Tokens',
maxTokensMessage:'Max Tokens is required',
maxTokensTip:
'This sets the maximum length of the model’s output, measured in the number of tokens (words or pieces of words).',
