Pretrained Generative Language Models as General Learning Frameworks for Sequence-Based Tasks (2402.05616v1)

Abstract: We propose that small pretrained foundational generative LLMs with millions of parameters can be utilized as a general learning framework for sequence-based tasks. Our proposal overcomes the computational resource, skill set, and timeline challenges associated with training neural networks and LLMs from scratch. Further, our approach focuses on creating small and highly specialized models that can accurately execute a challenging task of which the base model is incapable of performing. We demonstrate that 125M, 350M, and 1.3B parameter pretrained foundational LLMs can be instruction fine-tuned with 10,000-to-1,000,000 instruction examples to achieve near state-of-the-art results on challenging cheminformatics tasks. We also demonstrate the role of successive LLM fine-tuning epochs on improved outcomes, as well as the importance of both data formatting and pretrained foundational LLM selection for instruction fine-tuning success.