通过几何调控缓解大模型生成中的模式坍缩问题。
Escaping Mode Collapse in LLM Generation via Geometric Regulation
- 从动态系统视角将模式坍缩视为表征空间维度降低导致的轨迹受限。
- 提出轻量级在线干预方法RMR，使生成熵率低至0.8纳特/步仍稳定不坍缩。
- 适用于追求高多样性、低熵生成的场景，如长文本创作与逻辑推理。
模式坍缩是生成建模中的长期难题，在自回归文本生成中表现为显式循环、多样性渐失或轨迹过早收敛。本文从动力系统视角出发，将模式坍缩重新解释为由于几何坍缩导致的状态空间可达性下降：生成过程中模型内部轨迹被限制在表征空间的低维区域。这表明模式坍缩不仅是词元层面现象，无法仅通过符号约束或仅基于概率的解码启发式解决。基于此观点，我们提出强化模式调节（Reinforced Mode Regulation, RMR），一种轻量级、在线的状态空间干预方法，通过低秩阻尼对Transformer值缓存中的主导自增强方向进行调控。在多个大型语言模型上，RMR显著减少模式坍缩，并实现极低熵率下的稳定生成（最低达0.8 nats/step），而标准解码通常在约2.0 nats/step时即发生坍缩。
原文摘要 · Abstract (English)
Mode collapse is a persistent challenge in generative modeling and appears in autoregressive text generation as behaviors ranging from explicit looping to gradual loss of diversity and premature trajectory convergence. We take a dynamical-systems view and reinterpret mode collapse as reduced state-space accessibility caused by *geometric collapse*: during generation, the model's internal trajectory becomes confined to a low-dimensional region of its representation space. This implies mode collapse is not purely a token-level phenomenon and cannot be reliably solved by symbolic constraints or probability-only decoding heuristics. Guided by this perspective, we propose *Reinforced Mode Regulation* (RMR), a lightweight, online state-space intervention that regulates dominant self-reinforcing directions in the Transformer value cache (implemented as low-rank damping). Across multiple large language models, RMR substantially reduces mode collapse and enables stable generation at extremely low entropy rates (down to 0.8 nats/step), whereas standard decoding typically collapses near 2.0 nats/step.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。