mirror of
https://github.com/modelscope/DiffSynth-Studio.git
synced 2026-03-21 08:08:13 +00:00
update doc
This commit is contained in:
@@ -21,14 +21,16 @@
|
||||
* [FLUX](./Model_Details/FLUX.md)【TODO】
|
||||
* [Wan](./Model_Details/Wan.md)【TODO】
|
||||
|
||||
## Section 3: 特殊训练
|
||||
## Section 3: 训练框架
|
||||
|
||||
本节介绍 `DiffSynth-Studio` 所支持的特殊训练模式
|
||||
本节介绍 `DiffSynth-Studio` 中训练框架的设计思路,帮助开发者理解 Diffusion 模型训练算法的原理。
|
||||
|
||||
* FP8 训练
|
||||
* 端到端蒸馏训练
|
||||
* [Diffusion 模型基本原理](./Training/Understanding_Diffusion_models.md)
|
||||
* [标准监督训练](./Training/Supervised_Fine_Tuning.md)
|
||||
* [在训练中启用 FP8 精度](./Training/FP8_Precision.md)
|
||||
* [端到端的蒸馏加速训练](./Training/Direct_Distill.md)
|
||||
* 两阶段拆分训练
|
||||
* 差分 LoRA 训练
|
||||
* 拆分训练
|
||||
|
||||
## Section 4: 模型接入
|
||||
|
||||
|
||||
Reference in New Issue
Block a user