FBI-LLM: Scaling Up Fully Binarized LLMs from Scratch via Autoregressive Distillation
This research addresses the challenge of efficient and specialized hardware-friendly LLMs for AI deployment, presenting a novel approach rather than an incremental improvement.
This work tackles the problem of training fully binarized large language models from scratch to match the performance of full-precision counterparts, achieving competitive results in perplexity and task-specific effectiveness with models up to 7B parameters.
This work presents a Fully BInarized Large Language Model (FBI-LLM), demonstrating for the first time how to train a large-scale binary language model from scratch (not the partial binary or ternary LLM like BitNet b1.58) to match the performance of its full-precision counterparts (e.g., FP16 or BF16) in transformer-based LLMs. It achieves this by employing an autoregressive distillation (AD) loss with maintaining equivalent model dimensions (130M, 1.3B, 7B) and training data volume as regular LLM pretraining, while delivering competitive results in terms of perplexity and task-specific effectiveness. Intriguingly, by analyzing the training trajectory, we find that the pretrained weight is not necessary for training binarized LLMs from scratch. This research encourages a new computational framework and may facilitate the future design of specialized hardware tailored for fully 1-bit LLMs. We make all models, code, and training dataset fully accessible and transparent to support further research (Code: https://github.com/LiqunMa/FBI-LLM. Model: https://huggingface.co/LiqunMa/).