LAION-AI
make clere that answeres like " sorry i cant do that ..." are bad for the dataset.
make clear that answers like " sorry i cant do that ..." are bad for the datasets. many don't know this. you should just not answer them and mark the questions as "bad"
I think that is crucial. At the beginning, I was not aware that this explicitly not desired. I have just imitated what I just knew from ChatGPT and assumed that it's wanted to write texts like this.
The topic is certainly also generally connected with the general topic of guidelines. See also #1259 and #1295
yes but violence or other stuff is only to be supost to be marked with the matching option. and not with a answer like this. but i also though like you at first