DaoD/SPRING

[AAAI'25] SPRING: Learning Scalable and Pluggable Virtual Tokens for Retrieval-Augmented Large Language Models

25
/ 100
Experimental

This project helps developers enhance the factual accuracy and relevance of large language models (LLMs) for tasks like question-answering, without compromising the LLMs' original capabilities. It takes a pre-trained LLM and specialized 'virtual token embeddings' as input, then outputs an LLM that is better at incorporating retrieved information. This is for AI/ML engineers and researchers who are building applications that use Retrieval-Augmented Generation (RAG).

No commits in the last 6 months.

Use this if you need to significantly improve the performance of your LLMs in RAG scenarios, particularly for question-answering, while preserving their general generation abilities.

Not ideal if you are looking for a complete RAG system; this project focuses specifically on the LLM fine-tuning aspect using virtual tokens.

LLM development Retrieval-Augmented Generation NLP engineering AI research LLM fine-tuning
Stale 6m No Package No Dependents
Maintenance 2 / 25
Adoption 7 / 25
Maturity 16 / 25
Community 0 / 25

How are scores calculated?

Stars

26

Forks

Language

Python

License

MIT

Last pushed

Sep 24, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/rag/DaoD/SPRING"

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