search

tloen / alpaca-lora

Instruct-tune LLaMA on consumer hardware
Apache License 2.0
18.68k stars 2.22k forks source link

Finetune scenarios #624

Open Aekansh-Ak opened 8 months ago

Aekansh-Ak commented 8 months ago

I have around 2k rows of data coming in every hour in csv format.

  1. I wish to have a chatbot that can be queried upon this data, should I go for RAG or finetuning?
  2. If I go for finetuning will I have to train the model every time with new hourly data and for that how will I convert this csv file in to fine tune data format which is in json with Instruction, Input and output.
  3. Can I just finetune my data for instance on 1 week's data which will be 1000's of rows and later every hour I put the data in a vector database, pull in my finetuned model and then query vector database. i.e. Fine-tuning + RAG.

Your help is very much appreciated.