gmkim-ai/PromptKD

An official implementation of "PromptKD: Distilling Student-Friendly Knowledge for Generative Language Models via Prompt Tuning" (EMNLP 2024 Findings) in PyTorch.

35
/ 100
Emerging

This project helps machine learning engineers and researchers reduce the computational cost of deploying large language models. It takes an existing large, powerful language model (the "teacher") and a smaller, more efficient language model (the "student"), then applies a novel "prompt tuning" technique. The output is a smaller, fine-tuned student model that can perform complex generative tasks, like instruction-following, with performance comparable to the much larger teacher model but at a fraction of the inference cost.

No commits in the last 6 months.

Use this if you need to deploy a generative language model for tasks like instruction following, but are constrained by computational resources or latency, and want to achieve strong performance with a smaller model.

Not ideal if you are a business user looking for a no-code solution, or if you need to perform knowledge distillation for classification models rather than generative language models.

Large Language Models Model Compression Knowledge Distillation Generative AI Deployment AI Efficiency
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 5 / 25
Maturity 16 / 25
Community 14 / 25

How are scores calculated?

Stars

13

Forks

3

Language

Python

License

MIT

Last pushed

Nov 28, 2024

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/prompt-engineering/gmkim-ai/PromptKD"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.