Hastagaras/Gemmoy-9B-MK.I

gguf is here made available by Backyard AI

first try finetuning gemma 9b instruct. my data was a mess, i only noticed that after training, so this is just a test run

i formatted the dataset like below

<start_of_turn>user
{system}

{prompt}<end_of_turn>
<start_of_turn>model
{response}<end_of_turn>
<start_of_turn>user
{prompt}<end_of_turn>
<start_of_turn>model
{response}<end_of_turn>

and so on...

settings for st: context/instruct