gemma2:27b finetuning script and conversion to gguf format

#2
by ahsanmalik525 - opened

hi can you please share me how you fine tuned the 9b have you done that for gemma2 27b if not please share details or code files how I can do that.

Also will that script work with hermes function calling dataset as well.

Best Regards

Owner
edited Sep 3

Hey,

  • For fine-tuning, I used only LoRa with a small number of training epochs.
  • I will soon post a repository with Gemma-2-27B, including the GGUF format.
  • I did not use the hermes dataset in my training, and since my dataset is quite specific, I do not think you will be able to use it out of the box.
  • I used only my own dataset for training, which I prepared. I will also post it as soon as I finish working on the Russian version.
DiTy changed discussion status to closed

Sign up or log in to comment