20260625:很多新内容
This commit is contained in:
56
concepts/domain-aware-preference-optimization.md
Normal file
56
concepts/domain-aware-preference-optimization.md
Normal file
@@ -0,0 +1,56 @@
|
||||
---
|
||||
title: "Domain-Aware Preference Optimization"
|
||||
created: 2026-06-20
|
||||
updated: 2026-06-20
|
||||
type: concept
|
||||
tags: ["dpo", "preference-optimization", "domain", "lora", "post-training"]
|
||||
sources: ["https://arxiv.org/abs/2606.17800"]
|
||||
---
|
||||
|
||||
# Domain-Aware Preference Optimization (域感知偏好优化)
|
||||
|
||||
**Domain-Aware Preference Optimization** 是 [[maineCoon|MaineCoon]] 后训练的第一阶段:为不同社交视频域训练专门的 LoRA [[dpo|DPO]] expert。
|
||||
|
||||
## 为什么需要域感知
|
||||
|
||||
社交视频的质量标准因内容域而异:
|
||||
|
||||
| 域 | 质量重点 |
|
||||
|----|---------|
|
||||
| **Far Shot** | 全身结构稳定性、场景一致性 |
|
||||
| **Multi-Person Dialogue** | 说话人身份一致、轮流发言 |
|
||||
| **Motion** | 大幅度、时序连贯的身体运动 |
|
||||
| **Animation** | 风格一致的非写实渲染 |
|
||||
| **Dance** | 复杂肢体动作 + 音乐节奏同步 |
|
||||
|
||||
直接在所有目标上优化单一模型会引入**冲突偏好信号**。
|
||||
|
||||
## 方法
|
||||
|
||||
### Domain Preference Pairs
|
||||
对每个域 `d`:
|
||||
1. 用域质量过滤器选择高质量真实视频作为 `x⁺`
|
||||
2. 用当前 generator 生成同 prompt 的 `x⁻`
|
||||
3. 周期性用最新域模型刷新 `x⁻`,使偏好数据反映当前 failure modes
|
||||
|
||||
### Domain-Specialized DPO Experts
|
||||
从 native streaming checkpoint `θ₀` 出发,为每个域训练 LoRA adapter:
|
||||
```
|
||||
φ_d = θ₀ + Δ_d
|
||||
```
|
||||
使用 doubled-sequence interface(与 native training 相同),preferred 和 dispreferred 共享 prompt 和 noise,仅历史不同。
|
||||
|
||||
DPO loss:
|
||||
```
|
||||
L_DPO = -E[log σ(β_d · (ℓ⁻_φ - ℓ⁺_φ - ℓ⁻_θ₀ + ℓ⁺_θ₀))]
|
||||
```
|
||||
保留少量 reconstruction loss 在 preferred 样本上。
|
||||
|
||||
## 与 ROPD 的关系
|
||||
|
||||
域专家训练完成后,通过 [[reinforced-online-policy-distillation|ROPD]] 合并为单一部署策略。推理时**无需任何 domain adapter**。
|
||||
|
||||
## 参考
|
||||
- [[maineCoon|MaineCoon 论文]] Section 3.3
|
||||
- [[reinforced-online-policy-distillation|ROPD]]
|
||||
- [[dpo|Direct Preference Optimization]]
|
||||
Reference in New Issue
Block a user