EvoMerge: Neuroevolution for Large Language Models

Yushu Jiang

EvoMerge: Neuroevolution for Large Language Models

Neural and Evolutionary Computing 2024-02-02 v1 Artificial Intelligence Computation and Language Machine Learning

Authors: Yushu Jiang

Abstract

Extensive fine-tuning on Large Language Models does not always yield better results. Oftentimes, models tend to get better at imitating one form of data without gaining greater reasoning ability and may even end up losing some intelligence. Here I introduce EvoMerge, a systematic approach to large language model training and merging. Leveraging model merging for weight crossover and fine-tuning for weight mutation, EvoMerge establishes an evolutionary process aimed at pushing models beyond the limits of conventional fine-tuning.

Keywords

evolutionary optimization language modeling parameter-efficient fine-tuning

Cite

@article{arxiv.2402.00070,
  title  = {EvoMerge: Neuroevolution for Large Language Models},
  author = {Yushu Jiang},
  journal= {arXiv preprint arXiv:2402.00070},
  year   = {2024}
}

Comments

The current submission is the first draft, published for the sole purpose of sharing an idea and encouraging community effort. A more consolidated version may come later

EvoMerge: Neuroevolution for Large Language Models

Abstract

Keywords

Cite

Comments

Related papers