الملخص
Self-evolution lets large language models (LLMs) improve iteratively using their own generated data, but often suffers from self-evolution degeneration: performance improves, plateaus, then declines. Existing methods address this issue at the component level, targeting either the Questioner or the Solver, and overlook that self-evolution is a tightly coupled system. We propose a holistic framework based on learnable information gain, which measures how much novel, parameterizable information a round provides relative to the previous round. Theoretically, this gain equals the Kullback-Leibler divergence between the two rounds' data distributions plus their entropy change. Practically, it is estimated by fitting a small language model to the previous round and scoring new data via negative log-likelihood. Based on this diagnostic, we propose ATRI (Adaptive Training Regulation via Information-gain), which reweights samples within a round and halts training across rounds when information gain remains low. Experiments on popular datasets demonstrate the superiority of our proposal.
الكلمات المفتاحية
الموضوع
بيانات النشر
- المجلة
- غير متاح
- وصول مفتوح
- وصول مفتوح أخضر
اقتبس هذه المقالة
APA 7
Wang, C., Li, C., Shi, X., Liu, S., Wu, K. Y., Luo, Z., Zhang, S., Li, C., & Zhang, L. (2026). When Updating Stops Being Learning: Rethinking LLM Self-Evolution via learnable information gain. https://omanscience.com/ar/articles/when-updating-stops-being-learning-rethinking-llm-self-evolution-via-learnable-information-gain
MLA 9
Wang, Chenxu, et al. "When Updating Stops Being Learning: Rethinking LLM Self-Evolution via learnable information gain." https://omanscience.com/ar/articles/when-updating-stops-being-learning-rethinking-llm-self-evolution-via-learnable-information-gain.
شيكاغو (المؤلف–التاريخ)
Wang, Chenxu, Chaozhuo Li, Xinze Shi, Songyang Liu, Kyrie You Wu, Ziluowen Luo, Shun Zhang, Chenxi Li, and Litian Zhang. 2026. "When Updating Stops Being Learning: Rethinking LLM Self-Evolution via learnable information gain." https://omanscience.com/ar/articles/when-updating-stops-being-learning-rethinking-llm-self-evolution-via-learnable-information-gain.
هارفارد
Wang, C., Li, C., Shi, X., Liu, S., Wu, K. Y., Luo, Z., Zhang, S., Li, C. and Zhang, L. (2026) 'When Updating Stops Being Learning: Rethinking LLM Self-Evolution via learnable information gain', Available at: https://omanscience.com/ar/articles/when-updating-stops-being-learning-rethinking-llm-self-evolution-via-learnable-information-gain.
فانكوفر
Wang C, Li C, Shi X, Liu S, Wu KY, Luo Z, et al. When Updating Stops Being Learning: Rethinking LLM Self-Evolution via learnable information gain. https://omanscience.com/ar/articles/when-updating-stops-being-learning-rethinking-llm-self-evolution-via-learnable-information-gain
IEEE
C. Wang, C. Li, X. Shi, S. Liu, K. Y. Wu, Z. Luo, S. Zhang, C. Li, and L. Zhang, "When Updating Stops Being Learning: Rethinking LLM Self-Evolution via learnable information gain," https://omanscience.com/ar/articles/when-updating-stops-being-learning-rethinking-llm-self-evolution-via-learnable-information-gain.