gptq
Post-training 4-bit quantization for LLMs with minimal accuracy loss.
What this skill does for you
Post-training 4-bit quantization for LLMs with minimal accuracy loss. Use for deploying large models (70B, 405B) on consumer GPUs, when you need 4× memory reduction with <2% perplexity degradation, or for faster inference (3-4× speedup) vs FP16. Integrates with transformers and PEFT for QLoRA fine-tuning.
Quoted from the skill description in the pinned sourceWho made it
Use it with Edge
Set up Edge for me: read getedge.cc/SKILL.md and follow it. Then use Edge to load the gptq skill from davila7/claude-code-templates.
Paste it into Claude, ChatGPT, Codex or Cursor.
Or read the skill source first.
Security evidence
| Scanner | Date | Result | Findings |
|---|---|---|---|
| Edge static checks | 2026-09-27 | Pass | 0 |
| Cisco skill-scanner | 2026-09-27 | Pass | 0 |
| Semgrep Edge rules | 2026-09-27 | Pass | 0 |
Static scans check the code, not how well the skill works.
Scanner scope, raw findings and mirrored provider records
Edge static checks (Edge-run): edge-static/1.0.1. Scope: Static patterns for remote execution, credentials with network sends, obfuscation, prompt overrides, hidden Unicode, binaries and persistence. Revision: 6af7cdf57b810aca5e544a0f695bf058f1c2a744. Raw findings: . Counted findings: .
Cisco skill-scanner (Edge-run): cisco-skill-scanner/2.1.0. Scope: Local static, YARA, pipeline and behavioral analyzers; no LLM or AI Defense analysis. Revision: 6af7cdf57b810aca5e544a0f695bf058f1c2a744. Raw findings: 4. Counted findings: .
Semgrep Edge rules (Edge-run): semgrep/1.178.0+edge-rules.c15016a36fd3+offline-v1. Scope: Six Edge-authored static rules with Semgrep CE; offline execution and metrics disabled. Semgrep-maintained rules excluded. Revision: 6af7cdf57b810aca5e544a0f695bf058f1c2a744. Raw findings: . Counted findings: .
Source and licence
- Repository
- https://github.com/davila7/claude-code-templates
- Pinned skill file
- View source at revision
- Revision
- 6af7cdf57b810aca5e544a0f695bf058f1c2a744
- Package hash
- sha256:c4abfd6c71b0d8cf8aba585cf812e84b3b6e8726c602baffccbacc9ed58a798d
- Licence
- MIT