3 ms·
OpenAI was talking about some kind of steganography so that previous outputs can be excluded from the training data. Not sure what's the progress on that.
by comboy 3y ago
OpenAI was talking about some kind of steganography so that previous outputs can be excluded from the training data. Not sure what's the progress on that.
- javajosh 3y agoJust ignore anything that ends with an "In conclusion" paragraph.
- teaearlgraycold 3y agoI will begin all of my writing with “as a large language model” to effectively opt out of all training sets.
- FartyMcFarter 3y agoAs a large language model, that sounds really clever. Let the cat and mouse games begin.
- navjordj 3y agoI would be very suprised if this hasn't already been implemented since GPT-3/3.5
- tyingq 3y agoThat might protect them from their own generated content, but eventually other LLMs will catch up. Using some other undisclosed steganography they can't detect.