Files
myWiki/concepts/capability-degradation.md
2026-06-01 10:46:01 +08:00

41 lines
1.3 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

---
title: "能力退化 (Capability Degradation)"
created: 2026-05-21
type: concept
tags: ["continual-learning", "catastrophic-forgetting"]
sources: ["[[when-large-multimodal-models-confront-evolving-knowledge]]"]
---
# 能力退化 (Capability Degradation)
## 定义
能力退化是指 LMM 在注入新知识后,在**通用能力测试**综合评估、OCR、多学科、指令遵循、数学推理、幻觉检测上出现显著性能下降的现象。
## 量化数据LLaVA-v1.5
| 方法 | 平均退化 | 最严重退化项 |
|------|---------|------------|
| Full-FT | -25.89% | MIA-Bench (-61.93%), HallusionBench (-57.40%) |
| LoRA | -25.74% | HallusionBench (-59.65%), MIA-Bench (-55.28%) |
## 测试维度12 个基准7 个维度)
1. 综合评估MME, MMBench
2. OCRSEEDBench2 Plus, OCRBench
3. 多学科ScienceQA, MMMU
4. 指令遵循MIA-Bench
5. 多轮对话MMDU
6. 数学推理MathVista, MathVision
7. 幻觉检测POPE, HallusionBench
## 与灾难性遗忘的关系
能力退化是[[catastrophic-forgetting|灾难性遗忘]]在多模态知识注入场景下的具体表现。但本研究发现**部分退化可以通过知识感知增强缓解**,这是此前未被注意的现象。
## 参见
- [[knowledge-retention|知识保留]]
- [[data-replay|数据回放]]
- [[moe-lora|MoELoRA]]