Researcher urges papers to show actual training data samples, cites his CVPR24 practice
gabriberton · x · 2026-09-20
A researcher argues that a model is only as good as the data it sees, so every paper should include examples of the input data (post-augmentation samples actually fed to the model) in its supplementary material. He says he has adopted this practice in his own papers, sharing example tuples from a CVPR24 paper, and calls for qualitative training-data examples to become standard.
More from Research
- EcoLocator simulations show strong performance in continuous and discrete spaces — pastramimachine · 2026-09-20
- Nature paper: mammalian brain forms from two structures—hive minds may be real — juanbenet · 2026-09-20
- EcoLocator: deep learning infers climate-of-origin and location from genetic data — pastramimachine · 2026-09-20
- EcoLocator: deep learning predicts geographic origin and climate from genotypes — pastramimachine · 2026-09-20
- Vernor Vinge's 1993 "Technological Singularity" essay: superhuman AI within 30 years — akbirthko · 2026-09-20
- Self-Rewarding LLMs isn't news: RLAIF has long been standard practice — burny_tech · 2026-09-20