Spike Codec & Loss Pack

Community

AMP-safe spike codec & loss pack for SNNs.

Authorsovr610
Version1.0.0
Installs0

System Documentation

What problem does it solve?

Provides a modular, batched spike-encoding/decoding framework and a configurable spike-loss pack for training spike-based neural networks with strict AMP safety and diagnostics.

Core Features & Use Cases

  • Batch-first spike codec: encode inputs into SpikeBatch (B, T, N) and decode with AMP-hardening loss terms.
  • Loss composition: combine ProbSpikes, SpikeRateRegularization, TemporalConsistency, ISIRegularization, MembraneRegularization via SNNLossComposer with per-term diagnostics.
  • Deterministic/evaluation paths and generator control for reproducible experiments.
  • Reference templates and reference implementations to accelerate development, testing, and CI.

Quick Start

Run the AMP-stress tests to validate spike-codec losses across configurations and ensure end-to-end gradient flow.

Dependency Matrix

Required Modules

torch

Components

scriptsreferencesassets

💻 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: Spike Codec & Loss Pack
Download link: https://github.com/sovr610/refffiy/archive/main.zip#spike-codec-loss-pack

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 471,000+ vetted skills library on demand.