Recently, our research team at Shanghai Jiao Tong University and Tsinghua University released the first version of rLLM (relationLLM). The goal is to facilitate the machine learning workflows on those data stored in the relational databases with Large Language Models (LLMs).
As shown above, rLLM performs two key functions:
- Breaking down state-of-the-art GNNs, LLMs, and TNNs as standardized modules.
- Enabling the construction of novel models in a “combine, align, and co-train” way.
To learn more:
- Paper (Slides): https://arxiv.org/abs/2407.20157.
- Code: https://github.com/rllm-project/rllm.