Hi! I am very new to AI. Currently I have a service which generates logs as and now be it information or error. Incase of error, my project should provide me suggestions to solve the error.
Basically logs contain lot of unnecessary and necessary stuff containing error codes, exceptions or simple informations of the current service behaviour.
Some of the errors solution, can be found in the internet while some are very specific to service related errors.
I was thinking gen AI could help here where in I train a model with the service logs since it's kind of similar to NLP. And based on the context, it find errors. On a very high level. But I hear from couple of folks about training a llama model which already has wordly knowledge and training that model with the service data logs and running a azure databricks job that checks for the error logs and sends the context to the model for suggestions. Some mentioned about using vector database. But I am not sure how accure the suggestions would be.
So, I am really confused on how to even proceed about solving this problem....Any help or documentation would be of immense help to me...thank you!
[link] [comments]