The community member is using the GPTSimpleVectorIndex to index content, but the index.query is consuming a large number of tokens, making it costly to ask questions. The first comment suggests customizing the chunk size, which has reduced the token usage a bit. The second comment recommends trying the SentenceEmbeddingOptimizer during query-time as another optimization. There is no explicitly marked answer, but the community members are discussing ways to optimize the token usage for their queries.
Hello, quick qn, index.query is taking huge number of tokens, for each index.query call https://platform.openai.com/account/usage shows 4 requests with total of ~5000+ tokens, it is getting too costly to ask a question 😦 , I am using GPTSimpleVectorIndex to index the content. Any suggestions ?
