/vt/ - Virtual Youtubers

>>48543818
Most that we have is the oaicards guide, moth guide and our own personal testing (which doesn't mean much, since we all probably prompt our stress tests differently). Nothing comprehensive, the aforementioned two show you how they look, and make arguments for each. What makes a comprehensive test for outputs difficult in practice of each format is it may very well not be the format that's causing stress test oddities, but rather the words you used in the card (contradicting self, word salad that confused the AI), or natural quirks between the options we have.
Would it be worth trying eventually? Yeah honestly, we have like 4/5 diff formats+ running around. Same 4-5~ prompts, different formatting, 10-15 swipes each prompt, see if anything glaringly changes aside from token usage.