Unnatural Languages Are Not Bugs but Features for LLMs
This addresses the problem of understanding and leveraging unnatural languages for LLM robustness and performance, though it is incremental in reinterpreting existing phenomena.
The paper challenges the view that unnatural languages are bugs in LLMs, showing they contain latent features usable across models and tasks, with fine-tuned models achieving 49.71 win rates in Length-controlled AlpacaEval 2.0 on average.
Large Language Models (LLMs) have been observed to process non-human-readable text sequences, such as jailbreak prompts, often viewed as a bug for aligned LLMs. In this work, we present a systematic investigation challenging this perception, demonstrating that unnatural languages - strings that appear incomprehensible to humans but maintain semantic meanings for LLMs - contain latent features usable by models. Notably, unnatural languages possess latent features that can be generalized across different models and tasks during inference. Furthermore, models fine-tuned on unnatural versions of instruction datasets perform on-par with those trained on natural language, achieving 49.71 win rates in Length-controlled AlpacaEval 2.0 in average across various base models. In addition, through comprehensive analysis, we demonstrate that LLMs process unnatural languages by filtering noise and inferring contextual meaning from filtered words.