davila7 avatar

Knowledge Distillation

This is a template for compressing large language models down to smaller ones while keeping most of their performance. The techniques here (temperature scaling, soft targets, logit distillation) let you take a 70B parameter model and shrink it to 7B while retaining over 90% accuracy, which is huge for deployment costs. It's based on solid research, referencing the original Hinton paper from 2015 and newer work like MiniLLM. Honestly most useful if you're trying to move from expensive proprietary models to something you can actually run in production, or if you need to create a specialized model for a specific domain without training from scratch. The 478 installs suggest people are actively experimenting with this.

skills.sh installs
497
GitHub stars
33.2K
last updated
Aug 21, 2026

Install

$ npx -y skills add davila7/claude-code-templates --skill knowledge-distillation --agent claude-code

Installs into .claude/skills of the current project.

Files

This is a template for compressing large language models down to smaller ones while keeping most of their performance. The techniques here (temperature scaling, soft targets, logit distillation) let you take a 70B parameter model and shrink it to 7B while retaining over 90% accuracy, which is huge for deployment costs. It's based on solid research, referencing the original Hinton paper from 2015 and newer work like MiniLLM. Honestly most useful if you're trying to move from expensive proprietary models to something you can actually run in production, or if you need to create a specialized model for a specific domain without training from scratch. The 478 installs suggest people are actively experimenting with this.

Browse the repository on GitHub

Sponsors

Details

Repository
github.com/davila7/claude-code-templates
Category
Debugging
First seen
Jun 3, 2026

More from davila7/claude-code-templates

All 618

Recommended

More Debugging
1longbridge avatarlongbridge-risk-analysislongbridge/skillsRisk measurement and stress testing via Longbridge — computes VaR (historical simulation / parametric), CVaR (expected shortfall), max drawdown, Sharpe ratio, C496skills.sh installs
2aj-geddes avatarapi-paginationaj-geddes/useful-ai-promptsImplement efficient pagination strategies for large datasets using offset/limit, cursor-based, and keyset pagination. Use when returning collections, managing l495skills.sh installs
3longbridge avatarlongbridge-company-tearsheetlongbridge/skillsCompany tear sheet / one-pager via Longbridge Securities — generates a high-density 1–2 page company snapshot: business overview, key financials (revenue / net 495skills.sh installs
4longbridge avatarlongbridge-profit-analysislongbridge/skillsAccount-level profit and loss analysis via Longbridge Securities — simple return, time-weighted return (TWR), per-symbol P&L breakdown, and P&L by market. Suppo494skills.sh installs
5longbridge avatarlongbridge-post-investmentlongbridge/skillsPost-investment monitoring via Longbridge Securities — tracks portfolio holdings vs plan on a quarterly / monthly basis, extracts key KPIs (revenue growth / gro493skills.sh installs
6davila7 avatarautonomous-agentsdavila7/claude-code-templatesAutonomous agents are AI systems that can independently decompose goals, plan actions, execute tools, and self-correct without constant human guidance. The chal492skills.sh installs