📌 Basic Info
All the training data comes from images I generated using a style I really loved that I rolled from gpt-image-2.
Train model: minimax h3 image pruned bf16
Text Encoder: qwen3vl_32b_minimax_h3_int8_convrot
Recommend Output Size: >= 8MP
📝 Version Notes
V1✨
This is a test version
dataset: 23p
optimizer: adamw
lr: 1e-4
lr_scheduler: cosine
rank: 16
alpha: 16
steps: 805

