兼容Kohaku XL系列lora




我正在尝试训练一个用于动画背景的模型
希望在将来提升室内及室外复杂环境下的出图效果
I am trying to train a model for an animated background
We hope to improve the effect of drawing in complex indoor and outdoor environments in the future
我对0.3版本的训练集进行了大量优化,在0.1版本上进行了恢复,训练集从1w张提升到了9w张,泛化性有显著提升,同时加入了3k的自然语料训练集。
这只是一个过渡版本,我将继续对这个版本持续微调。
0.3正式版将在这周发布,希望大家积极返图。
I optimized the training set of version 0.3 a lot and restored it on version 0.1. The training set was increased from 1w pieces to 9w pieces, which significantly improved the generalization. At the same time, 3k natural corpus training set was added.
This is only a transitional release, and I will continue to fine-tune this release.
0.3 official version will be released this week, I hope everyone actively return to the picture.
兼容Kohaku XL系列lora
1. 轉载模型僅供學習與交流分享,其版權及最終解释權归原作者。
2. 模型原作者如需認領模型,請通過官方渠道联系海藝AI工作人員進行認證。我們致力於保護每一位創作者的權益。 點擊去認領
