model-extraction-relu-logits

Community

Reverse-engineer ReLU network weights.

AuthorZurybr
Version1.0.0
Installs0

System Documentation

What problem does it solve?

This Skill addresses the challenge of recovering the internal weight matrices of black-box ReLU neural networks when only input-output query access is available, crucial for understanding model behavior and security.

Core Features & Use Cases

  • Weight Matrix Recovery: Extracts the first layer weight matrix (A1) of a two-layer ReLU network (f(x) = A2 @ ReLU(A1 @ x)).
  • Model Extraction Attacks: Enables reverse-engineering of neural network parameters for security analysis or model understanding.
  • Use Case: A security researcher wants to understand the internal structure of a deployed neural network model without access to its training data or architecture. This skill can help recover the hidden layer's weight matrix by querying the model.

Quick Start

Use the model-extraction-relu-logits skill to extract the weight matrix A1 from the provided black-box ReLU network function.

Dependency Matrix

Required Modules

None required

Components

references

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: model-extraction-relu-logits
Download link: https://github.com/Zurybr/lefarma-skills/archive/main.zip#model-extraction-relu-logits

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 223,000+ vetted skills library on demand.