a lot of variables that once examined by Claude Code and gemini 3 pro themselves, it can be safely said that it's 错漏百出.
im really tired but i want this to work correctly. the derived features must be scientifically sound, biologically backed, and statistically accurate. everything else will result in bad training data for the ENR.
this is tough. i should be the final guardrail but im not proficient enough in image to numerical feature extraction in terms of microscopy images. i can't even normalize dna stuff myself. well technically i know about z score standardization but that's my max limit of knowledge... for now.
i need to find time to learn, but there is not much time... there might be more after summer break starts tho, as imaging (revealed by b to me) that it will continue for not 100, but 120 days max before staining.
currently after multiple prompts that require triple checks on both gemini and claude opus derives a guardrail to one final checkpoint for the potential problem of anisotropy.
anything else? not sure. and my instinct tells me something might be fucked by them. so i have to prob dig deeper later into the data extraction schema. to be more exact, february 21st, is when i get highly technical in this specific thing.