data source?
see title, whered it come from?
HI, story data was batch download from different sites from 2018 to 2022 many of those site now offline i not remember all sites , i download for my personal offline archive not for ai, archive size 4.8GB compressed, 16GB uncompresed, it count more than 3500000 total of lewd stories in english, in 2023 i use this small dataset for train lora for llama2 using llama.cpp finetune tool, remade for llama 3 ,also i share big datasets here, named yiff-extreme 2k stories, and yiffcomebrain 10k stories datasets
i found my dataset in you repo, it's ok, but add link to me , because it was semi-private sharing
https://huggingface.co/datasets/Fizzarolli/yiff-extreme
i found my dataset in you repo, it's ok, but add link to me
hm? i got it from koboldai, and unless you're the same person who sent the mega link there, then we evidently just got the same data from difference sources
i m not using discord and mega lol , i share original archive here, my old account with same name was hijacked and deleted but i have link history, here is proof that i m creator of dataset, the post number 3 https://huggingface.co/jeiku/Salami_3B/discussions/1 also contain unformated data! here the link https://pixeldrain.com/u/cEjXikRo
KatyTheCutie mentioned my name in posts
well, i got it from katy, sooo...
raw data was shared by me on Feb 2 in this https://huggingface.co/jeiku/Salami_3B/discussions/1, but the katy post is in march ,one month after i share it here lol.
@Fizzarolli
do you want use this data set?
i try fine tune this data set on llama 3.1-instruct using unslothon colab, sad every time get error, tryed to reformat it to fix error but no luck,colab url https://colab.research.google.com/drive/1Ys44kVvmeZtnICzWz0xgpRnrIOjZAuxp
i did a couple models with it in the mix! check out all the ones named rosier on my profile