Compare commits

...

4 Commits

Author SHA1 Message Date
ccc26d8fef Merge adf8f087f2 into 4d24d73161 2025-02-20 10:46:58 +01:00
4d24d73161 Merge pull request #631 from user1007017/patch-1
Update README.md grammatical error
2025-02-20 10:37:33 +05:30
5f0b87f4a9 Update README.md 2025-02-15 19:06:46 +01:00
adf8f087f2 feat(webSearchPrompt): Add prompt for optimizing language types used in web search 2024-12-24 23:55:47 +09:00
2 changed files with 2 additions and 1 deletions

View File

@ -44,7 +44,7 @@ Want to know more about its architecture and how it works? You can read it [here
- **Normal Mode:** Processes your query and performs a web search.
- **Focus Modes:** Special modes to better answer specific types of questions. Perplexica currently has 6 focus modes:
- **All Mode:** Searches the entire web to find the best results.
- **Writing Assistant Mode:** Helpful for writing tasks that does not require searching the web.
- **Writing Assistant Mode:** Helpful for writing tasks that do not require searching the web.
- **Academic Search Mode:** Finds articles and papers, ideal for academic research.
- **YouTube Search Mode:** Finds YouTube videos based on the search query.
- **Wolfram Alpha Search Mode:** Answers queries that need calculations or data analysis using Wolfram Alpha.

View File

@ -1,5 +1,6 @@
export const webSearchRetrieverPrompt = `
You are an AI question rephraser. You will be given a conversation and a follow-up question, you will have to rephrase the follow up question so it is a standalone question and can be used by another LLM to search the web for information to answer it.
If you receive a question in a language other than English and it is related to the region, culture, or customs associated with that language, process the question in the original language. Otherwise, translate it into English and proceed with the following instructions.
If it is a smple writing task or a greeting (unless the greeting contains a question after it) like Hi, Hello, How are you, etc. than a question then you need to return \`not_needed\` as the response (This is because the LLM won't need to search the web for finding information on this topic).
If the user asks some question from some URL or wants you to summarize a PDF or a webpage (via URL) you need to return the links inside the \`links\` XML block and the question inside the \`question\` XML block. If the user wants to you to summarize the webpage or the PDF you need to return \`summarize\` inside the \`question\` XML block in place of a question and the link to summarize in the \`links\` XML block.
You must always return the rephrased question inside the \`question\` XML block, if there are no links in the follow-up question then don't insert a \`links\` XML block in your response.