torchdebug

Community

Precision debugging for PyTorch models on TianShu hardware.

Authordongg622
Version1.0.0
Installs0

System Documentation

What problem does it solve?

This Skill enables users to identify and troubleshoot precision discrepancies in PyTorch models running on TianShu AI chips, facilitating migration and adaptation with detailed operator comparison.

Core Features & Use Cases

  • Operator-level accuracy comparison between TianShu, CUDA, and CPU outputs to locate deviations.
  • Layer-wise inspection including input/output tensor snapshots and error metrics.
  • HTML report generation that visualizes model differences for in-depth analysis.
  • Use Case: When a model trained on NVIDIA GPU exhibits accuracy issues after deployment on TianShu, this Skill helps pinpoint specific operators or layers causing the discrepancy.

Quick Start

Wrap your model with the torchdebug.Debugger and run the comparison to generate an HTML report of precision deviations.

Dependency Matrix

Required Modules

torchdebug

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: torchdebug
Download link: https://github.com/dongg622/china-ai-chip-skill/archive/main.zip#torchdebug

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.