IMC-CLINIC: Coupled Loss-Informed Newton Iterations for Clipping in Analog In-Memory Computing

📄 arXiv: 2609.35586v1 📥 PDF

作者: Yung-Chin Chen, Chia-Yu Chen, Naveen Verma

分类: cs.AI

发布日期: 2026-09-28

💡 一句话要点

提出IMC-CLINIC以解决模拟内存计算中的剪切优化问题

🎯 匹配领域: 支柱九：具身大模型 (Embodied Foundation Models)

关键词: 模拟内存计算 剪切优化 量化误差 牛顿法 大型语言模型 能效 校准方法

📋 核心要点

  1. 现有剪切方法未能有效优化IMC中的多重误差源，且通常依赖于昂贵的搜索校准，导致效率低下。
  2. IMC-CLINIC通过分析代理模型联合优化激活和权重剪切因子，采用保护的牛顿法进行高效校准。
  3. 在多个模型和数据集上，IMC-CLINIC的零-shot准确率提升了6.5-11.5个百分点，校准时间减少了10.0-12.1倍。

📝 摘要（中文）

模拟内存计算（IMC）为能效高的大型语言模型推理提供了有前景的解决方案，但其精度受限于模拟到数字转换器（ADC）的量化误差。现有的剪切方法未能有效优化IMC中的多重误差源，且通常依赖于昂贵的搜索校准。本文提出IMC-CLINIC，一个基于IMC矩阵乘法输出误差的分析代理的剪切校准框架，能够高效评估梯度和近似曲率。IMC-CLINIC通过保护的牛顿法联合优化激活和权重剪切因子，显著提高了多个模型和数据集的零-shot准确率，并减少了校准时间。

🔬 方法详解

问题定义：本文旨在解决模拟内存计算中由于ADC量化引起的输出误差问题。现有方法未能有效优化激活和权重剪切因子，导致性能下降和校准效率低下。

核心思路：IMC-CLINIC通过构建一个分析代理模型，联合考虑操作数量化、剪切引起的偏差和ADC量化，从而高效评估输出误差的梯度和近似曲率。

技术框架：IMC-CLINIC的整体架构包括数据采集、分析代理模型构建、梯度计算和牛顿法优化四个主要模块。首先，通过小规模的校准集获取数据，然后构建代理模型，最后使用牛顿法优化剪切因子。

关键创新：IMC-CLINIC的主要创新在于其分析代理模型能够同时处理多种误差源，并通过保护的牛顿法实现高效的剪切因子优化，这与传统的数字量化剪切方法有本质区别。

关键设计：在设计中，IMC-CLINIC使用了特定的损失函数来量化输出误差，并通过小规模的校准集来提高优化效率，确保优化器在全局最优解附近的1%以内。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

IMC-CLINIC在多个模型和数据集上实现了6.5-11.5个百分点的零-shot准确率提升，相较于基线的网格搜索方法，校准时间减少了10.0-12.1倍，显示出显著的效率和准确性提升。

🎯 应用场景

IMC-CLINIC在大型语言模型推理中具有广泛的应用潜力，尤其是在需要高能效和高准确度的场景中。其优化方法能够有效降低校准时间，提高模型的推理性能，适用于各种基于模拟内存计算的应用，如智能助手和自然语言处理任务。

📄 摘要（原文）

Analog in-memory computing (IMC) offers a promising path toward energy-efficient large language model (LLM) inference by executing matrix multiplications (MatMul) directly within memory arrays in the analog domain. Its efficiency, however, comes with an additional source of error: limited-precision analog-to-digital converters (ADCs) quantize accumulated analog partial sums, introducing output-side error distinct from conventional activation and weight quantization at the MatMul inputs. Clipping can mitigate both operand and ADC quantization errors, but the optimal clipping factors must jointly balance activation rounding and clipping, weight rounding and clipping, and ADC quantization. Existing clipping methods, designed for digital quantization, do not explicitly optimize these coupled sources of IMC error and often rely on costly search-based calibration. We introduce IMC-CLINIC (Coupled Loss-Informed Newton Iterations for Clipping), a clipping calibration framework based on an analytical surrogate for IMC MatMul output error. The surrogate jointly models operand quantization, accumulated clipping-induced bias, and ADC quantization, enabling efficient evaluation of its gradient and approximate curvature from a small calibration set. IMC-CLINIC jointly optimizes activation and weight clipping factors using a safeguarded Newton-type method. Across multiple models and datasets, it improves average zero-shot accuracy by 6.5-11.5 percentage points over the grid search baseline while reducing calibration time by factors of 10.0-12.1. Its analytical surrogate closely tracks empirical IMC output error, and its optimizer is certified within 1% of the global optimum under the loss objective across all projections on two representative models.