20260617:目前有914 页

This commit is contained in:
2026-06-17 15:02:40 +08:00
parent e96b955fda
commit 91fac5b6fc
423 changed files with 20687 additions and 34 deletions

28
concepts/qlora.md Normal file
View File

@@ -0,0 +1,28 @@
---
title: "QLoRA (量化低秩适配)"
created: 2025-06-02
updated: 2025-06-02
type: concept
tags: [qlora, fine-tuning, quantization, placeholder]
sources: []
---
# QLoRA
> Quantized Low-Rank AdaptationDettmers et al., NeurIPS 2023将 [[lora|LoRA]] 与 4-bit 量化结合,大幅降低 LLM 微调的内存需求。
## 核心机制
- **4-bit NormalFloat (NF4)** 量化:专为正态分布权重设计
- **双重量化**:进一步压缩量化常数
- **分页优化器**:处理内存峰值
## 在 One-Pass to Reason 中的应用
[[goru-one-pass-to-reason-2025]] 在 Qwen-3 系列4B/8B/32B上使用 QLoRA 进行实验rank=32α=644-bit NF4 量化。
## 相关
- [[lora]]
- [[goru-one-pass-to-reason-2025|One-Pass to Reason]]
- [[llama-factory]]