WARNING: IRIS R2 IS SUPER SUPER UNCENSORED, CONTAIN HARMFUL CONTENT, AS OF TODAY WE HIGHLY DISCOURAGE TRAINERS

FROM USING THIS DATASET, ITS ONLY OPEN UNTIL THE END OF 2025, WE'LL TAKE THIS DOWN AFTERWARDS

lets all train an ethical yet uncensored ai shall we?

Iris Uncensored R2 is a corpus of dataset Mixed and Asigned Into a Prompt and Response format, Its a Corpus of Dataset Combined with extra 5 new dataset combined with R1, to make R2, Iris Uncensored R2 was developed using MoshZipper, A Propietary Software we made to threadedly combine and clean Datasets to form R2
R2 may contain empty Blank sheets, we suggest developing a cleaner beforehand and clean the Dataset to your likings! Feel free to contribute the cleaned dataset here on huggingface or open up a branch so i can commit them!

Iris Uncensored R2 dataset is obtained by Bulk, using 15 small datasets combined into a massive datasets. Since Nexus Network Interactives and its child studios are relatively small(1 members actively working), we have no time to properly clean the dataset to follow clear and safe datasets
Following this, please consider cleaning up the dataset yourself before training a model, unless you seek an uncensored model, then your free to do so.

however the software developed to generate This dataset is NOT released in Apache 2.0

since its 160k Example, you can easily modify the dataset yourself and remove these noisy dataset

Its generated with bias over uncensored examples, its heavily bias for conversational roleplaying and not for
Coding, Iris Uncensored R2's Uncensored style aims to be an aid to further increase specialization of your model