0426 CRef/SRef LoRA Triplet Dataset This dataset contains CRef/SRef LoRA triplets exported from the 0426 diffusion training data. Each training example has three images: content : content reference image, used as cref 0 style : style reference image, used as sref 0 target : image generated from the combined content + style condition Use triplets.csv as the main entry point. Image level CSV files are provided only for deduplicated metadata and provenance lookup. Sources Directory Base model Original source Triplets : cref sref/qwen/ qwen cref sref qwen lora part1 33,582 cref sref/flux/ flux cref sref flux lora part1 273,682 cref sref/illustrious/ illustrious cref sref illustrious lora part1 172,589 Layout How To Use Pick one source directory and read its triplets.csv : Which Prompt Fields Are Used For Training? The 0426 training config uses the three lora triplet sources: In the training loader, a sample is not represented by a single prompt string. Each training choice is: Only the final target image has require loss=True ; the two text fields are conditioning text. For these lora triplet sources, the training DB provides 8 text choices per sequence. Each choice uses exactly one in…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy