LLM for TimeSeries Data

AI Discussions
,
1

I would like to know about any existing LLM that can be used with TimeSeries Data for the following generic requirements:

  • The TimeSeries data can have numbers ranging from fractions to 10 digits of multiple features at that time.
  • User should be able to provide the new dynamic data as RAG
  • The prompts can be any type of query on the timeseries data. Some examples of the query types can be on prediction, statistics, general Q&A about the data. All these prompts should be answered by the LLM , using the new/dynamic data as RAG
  • The number of embeddings should be max 500 ( including the input prompt and RAG embeddings) , so the individual numbers should not be treated as individual embedding
  • We can use multiple LLM’s to satisfy each requirement individually, but we would like to know for any single LLM that can treat these lake of numbers like the way that text or Image based LLM’s treats the Text or Image, without increasing the number of embeddings, since each number is different .
2 Likes
2

Dear @AnanthM,

Welcome to the community.

Right now I do not have the answer of your question, but let you know soon.